Finalrep 1

Optimal Switching Strategy for Power System with FACTS
Based Controllers
A Thesis
Submitted to the
Visvesvaraya Technological University, Belagavi
For the degree of
Doctor of Philosophy
In
Electronics & Communication Engineering
By
Yathisha L
Under the guidance of
Dr. Sudarshan S Patil Kulkarni

Professor
Department of Electronics & Communication Engineering,

Sri Jayachamarajendra College of Engineering,
Mysuru-570006.
Dedicated
To
My Parents, Wife, Family Members

and Well-Wishers
****Yathisha L
SRI JAYACHAMARAJENDRA COLLEGE OF
ENGINEERING
DEPARTMENT OF ELECTRONICS & COMMUNICATION ENGINEERING
Mysore-570006.
CERTIFICATE
This is to certify that the thesis entitled “Optimal Switching Strategy for Power
System with FACTS Based Controllers,” submitted to the Vivesvaraya Technological
University, Belagavi by Mr.Yathisha L, USN No. 4JC11PEM03 for the award of Doctor of
Philosophy in Electrical Engineering, is a bonafide record of research work carried out by
him at Electronics & Communication Engineering Department, Sree Jayachmarajendra
College of Engineering, Mysuru under my supervision and guidance.
The candidate has fulfilled all the prescribed requirements.
The Thesis which is based on candidate’s own work, has not submitted elsewhere for a
degree/diploma to the best of my knowledge and belief.
In my opinion, the thesis is of standard required for the award of a Doctor of Philosophy
degree in Electronics & Communication Engineering.
Dr. Sudarshan S Patil Kulkarni

Professor
Electronics & Communication Department
SJCE, Mysuru
DECLARATION
I Yathisha L, certify that this thesis is the result of research work done by me under
the supervision of Dr. Sudarshan S Patil kulkarni at Electronics and Communication
Engineering Department, Sri Jayachamarajendra College of Engineering, Mysore, India.
I am submitting this thesis for possible award of Doctor of Philosophy (Ph.D.) degree in
Electronics & Communication Engineering for the Visvesvaraya Technological University,
Belagavi. I further certify that this thesis has not been submitted by me for award of any
other degree/diploma of this or any other University.
Date : Yathisha L
Place :Mysuru Research Scholar
ACKNOWLEDGEMENT
This thesis is a result of research that has been carried out at Electronics and
Communication Engineering Department, Sree Jayachamarajendra College of Engineer-
ing, Mysuru. During this period, I came across with a great number of people whose
contributions in various ways helped my field of research and they deserve special thanks.
It is a pleasure to convey my gratitude to all of them.
First and foremost, I would like to express my sense of gratitude and indebtedness
to my supervisor Dr. Sudarshan S Patil Kulkarni, Professor, Department of Electronics
and Communication Engineering, for his inspiring guidance, encouragement, and untiring
effort throughout the course of this work. His timely help and painstaking efforts made it
possible to present the work contained in this thesis. I consider myself fortunate to have
worked under his guidance. His professional attitude and great personality have made
period of research a very rewarding and memorable experience. His trust and guidance
inspired me in the most important moments of making right decisions. Finally, I am proud
to record that I had opportunity to work with an exceptionally experienced scientist like
him.
I am grateful to Dr. T Nagabushan, Principal and Dr. M N Shanmukhaswamy,

Professor & Head, Department of Electronics and Communication Engineering, for their
kind support and concern regarding my academic requirements.
I express my thankfulness to the faculty, co-scholars and staff members of Depart-

ment of Electronics and Communication Engineering, SJCE, Mysuru for their continuous
encouragement and suggestions.
I am highly grateful to the authorities of the Visvesvaraya Technological University,

Belagavi for giving me an opportunity to pursue my research studies in the University.
I express my heartfelt thanks to the international journal reviewers for giving their
valuable comments on the published papers in different international journals, which helps
to carry the research work in a right direction. I also thank to the international conference
organizers for intensely reviewing the published papers.
I am highly grateful and indebted to the Chairman, Secretary, Treasurer and

members of ATME College of Engineering, Mysuru for their constant encouragement and
excellent facilities extended to me in the institution.
I express my sincere thanks to Dr.L Basavaraj, Principal and Dr. Mahesh P

K, Professor & Head of the Department of Electronics and Communication Engineering,
ATMECE, Mysore, and all my colleagues and non-teaching staff of my department (Elec-
tronics & Communication Department, ATMECE, Mysore) supported me at every stage
of my research. I sincerely express my profound thanks to all of them.
I express my deep sense of gratitude and reverence to my beloved father Sri.

Lokesha R, mother Smt. Malathi K and my wife Smt. Shilpa T V who supported and
encouraged me all the time, no matter what difficulties I encountered. I would like to
express my greatest admiration to all my family members and relatives for their positive
encouragement that they showered on me throughout this research work. Without my
family’s sacrifice and support, this research work would not have been possible. It is a
great pleasure for me to acknowledge and express my appreciation to all my well wishers
for their understanding, relentless supports, and encouragement during my research work.
Last but not the least, I wish to express my sincere thanks to all those who helped me
directly or indirectly at various stages of this work.
Above all, I would like to thank The Almighty God for the wisdom and per-
severance that he has been bestowed upon me during this research work, and indeed,
throughout my life.
ABSTRACT
The rapid development of cities, industries, etc. alternately increases the power
demand. Today power system has been much loaded compared to earlier days. Due to
environmental & other economic reasons, it is difficult to build new power lines and to
reinforce the existing ones. The problem is to operate the power system for increased
loading in the available lines itself. The current challenge in today’s highly complex power
systems operating over range of operating conditions and disturbances is to stabilize the
system, controlling the damping of low frequency oscillations, minimizing the overshoots
for rotor angle & rotor speed deviations. This has been an area of intense research for
electrical engineers in past few decades. Power system industry has been using Power
System Stabilizers (PSS) and Flexible AC Transmission Control (FACTS) based devices
such as Unified Power Flow Controller (UPFC) for this purpose.
The current challenge for the control system community today is to obtain the
maximum benefit from PSS & UPFC devices for controlling, stabilizing, minimization of
overshoots, etc. for power system. The control system community has started inventing
new control techniques for PSS & UPFC for damping oscillations.
Following, the works of control system community for power system, focus of this
thesis is on implementing switched linear control theory concepts for the Single Machine
Infinite Bus (SMIB) linearized Phillips heffron power system model, initially in this thesis
an overview of switched linear control theory concepts with optimal control basic designs
is explained followed by power system linearization & modeling is described. Later, the
switched linear theory concepts are proposed for the current thesis in various novel switch-
ing techniques like firstly, switching between two optimal feedback controllers are proposed
for PSS & UPFC. Secondly, for the coordinated design of PSS & UPFC the switching
between two feedback controllers are designed. Thirdly, switching between different oper-
ating points based on three scenarios is developed for UPFC. Fourthly, switching between
various combinations of uncoordinated/coordinated PSS & UPFC control inputs are pro-
posed for the power system subjected to with & without disturbances and Finally, multi
stages of switching is developed for UPFC with first of switching between two optimal
feedback controllers followed by second stage of switching between two operating points
of UPFC.
The results of the investigations conducted in this thesis show that the achieved
optimal switched linear control designs are effective in damping low frequency oscillations
by minimizing the peak overshoots & settling time of state variables in the Phillips heffron
model. The switched linear concepts for optimal controllers design procedures adopted
in this thesis are general and can be applied to other FACTS devices incorporated in a
power system. The results and discussion presented in this thesis should provide valuable
information to electric power utilities engaged in planning and operating FACTS devices.
TABLE OF CONTENTS
LIST OF FIGURES vi
LIST OF TABLES x
1 INTRODUCTION 2
1.1 Literature Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1.1 Power System with FACTS Based Controllers . . . . . . . . . . . . 4
1.1.2 Switched Linear Systems . . . . . . . . . . . . . . . . . . . . . . . . 9
1.1.3 Switching Control Strategies for Power System . . . . . . . . . . . . 11
1.1.4 Drawbacks of the Existing Control Techniques for Power System . . 11
1.2 Objectives of the Research . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4 Scope, Limitations and Delimitation’s of the Research . . . . . . . . . . . . 14
1.5 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2 PRELIMINARIES 19
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 Power System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2.1 Linearization Concepts . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2.2 Power System Stability . . . . . . . . . . . . . . . . . . . . . . . . . 22
i
2.2.3 Phillips Heffron SMIB Model . . . . . . . . . . . . . . . . . . . . . 24
2.2.4 Flexible AC Transmission Sytems (FACTS) . . . . . . . . . . . . . 29
2.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3 OPTIMIZATION USING SWITCHING CONTROL 35
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.2 Optimal Control Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.2.1 Linear Quadratic Regulator Control . . . . . . . . . . . . . . . . . . 35
3.2.2 Linear Quadratic Guassian Control . . . . . . . . . . . . . . . . . . 38
3.3 Switched Linear Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3.1 Motivation for switching . . . . . . . . . . . . . . . . . . . . . . . . 41
3.3.2 Switching control Algorithms . . . . . . . . . . . . . . . . . . . . . 42
3.4 Optimization of Performance Index Via Switching Control . . . . . . . . . 46
3.4.1 Optimization of JSwitch Using Switching Algorithm I . . . . . . . . 47
3.4.2 Optimization of JSwitch Using Switching Algorithm II . . . . . . . . 48
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4 MODELING, OPTIMAL CONTROL DESIGN & SWITCHING CON-

TROL OPERATION OF PSS & UPFC 55
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2 Linearized Model of SMIB with PSS . . . . . . . . . . . . . . . . . . . . . 56
4.2.1 Experiments using Switching Control Algorithms . . . . . . . . . . 58
4.2.2 Result Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.3 Linearized model of SMIB with UPFC . . . . . . . . . . . . . . . . . . . . 64
ii
4.3.1 Experiments using Switching Control Algorithm . . . . . . . . . . . 66
4.3.2 Result Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.4 Switching Control Design for Simultaneous Coordinated Deign of PSS And
UPFC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.5 Preliminary Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.6 Experiments using switching control algorithm . . . . . . . . . . . . . . . . 79
4.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5 LQR AND LQG BASED OPTIMAL SWITCHING TECHNIQUES FOR

PSS & UPFC IN POWER SYSTEMS 86
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.2 Power system modelling with PSS & UPFC . . . . . . . . . . . . . . . . . 87
5.3 Switching Control Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.3.1 Switching Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.4 Experimental Set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.5 Comparison of simulation results for LQR & LQG optimal switching con-
trollers with disturbances . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.5.1 Scenario I: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.5.2 Scenario II: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
6 SWITCHING CONTROL TECHNIQUES VIA OPTIMIZATION FOR

UPFC WITH EFFECT TO OPERATING CONDITIONS IN POWER
SYSTEM 108
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
iii
6.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.3 Proposed switching control for different operating conditions . . . . . . . . 112
6.3.1 Switched Linear Control Systems . . . . . . . . . . . . . . . . . . . 112
6.3.2 Switching Control Techniques . . . . . . . . . . . . . . . . . . . . . 113
6.4 Simulation results and Comparison of switching techniques for all the pro-
posed scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
7 SWITCHING CONTROL OF MULTI- LQRs FOR UPFC AS APPLIED

TO LOADING CONDITIONS IN POWER SYSTEM 135
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
7.2 Preliminary Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
7.2.1 Bryson Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
7.2.2 Bouderal Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
7.2.3 Multistage Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
7.2.4 Simulation Results of Preliminary Analysis . . . . . . . . . . . . . . 140
7.2.5 Discussion on preliminary Analysis . . . . . . . . . . . . . . . . . . 141
7.3 Multi-LQRs Switching Control . . . . . . . . . . . . . . . . . . . . . . . . . 144
7.3.1 Switched Linear System . . . . . . . . . . . . . . . . . . . . . . . . 144
7.3.2 Switching Control Strategies . . . . . . . . . . . . . . . . . . . . . . 145
7.3.3 Simulation Results of Multi-LQR Switching . . . . . . . . . . . . . 146
7.3.4 Discussion on multi-LQR switching control strategy . . . . . . . . . 151
7.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
iv
8 CONCLUSION AND FUTURE WORK 156
8.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
8.2 Contributions of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
8.3 Recommendations for future work . . . . . . . . . . . . . . . . . . . . . . . 159
8.3.1 Switching between feedback controllers . . . . . . . . . . . . . . . . 159
8.3.2 Switching control for Coordinated PSS & UPFC . . . . . . . . . . . 159
8.3.3 Switching strategies with disturbance . . . . . . . . . . . . . . . . . 159
8.3.4 Optimal switching strategies: Loading conditions . . . . . . . . . . 160
8.3.5 Multi LQRs Switching . . . . . . . . . . . . . . . . . . . . . . . . . 160
8.3.6 General Recommendations . . . . . . . . . . . . . . . . . . . . . . . 160
v
LIST OF FIGURES
1.1 Multicontroller Architecture of Switched Systems . . . . . . . . . . 9
2.1 Power Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2 Overview of stability analysis . . . . . . . . . . . . . . . . . . . . . . . 24
2.3 General Linear Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4 AVR connected to Generator . . . . . . . . . . . . . . . . . . . . . . . 25
2.5 Phasor diagram of SMIB systems . . . . . . . . . . . . . . . . . . . . 26
2.6 Phillips heffron model (a) . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.7 Phillips heffron model (b) . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.8 Phillips heffron model (c) . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.9 Phillips heffron model (d) . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.10 Conventional Thyristor Based FACTS Controller . . . . . . . . . . 30
2.11 STATCOM voltage source . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.12 UPFC configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.13 FACTS applications year wise chart . . . . . . . . . . . . . . . . . . . 31
3.1 Block Diagram of Optimal LQG compensator . . . . . . . . . . . . . 39
4.1 Block diagram of IEEE excitation system . . . . . . . . . . . . . . . 56
4.2 Block diagram of power system installed with PSS . . . . . . . . . 56
4.3 Switched Linear system for PSS control input . . . . . . . . . . . . 61
vi
4.4 Rotor angle & Rotor speed deviations responses for individual
LQR & switching control algorithms. . . . . . . . . . . . . . . . . . . 62
4.5 Rotor angle & Rotor speed deviation respones for pole placement
control method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.6 Rotor angle & Rotor speed deviation switching signals responses. 63
4.7 Rotor angle & Rotor speed deviation switching signal responses. 63
4.8 Block Diagram of Power System installed with UPFC . . . . . . . 65
4.9 General linear switched system for UPFC . . . . . . . . . . . . . . . 67
4.10 Rotor speed deviation response for mE & δE . . . . . . . . . . . . . . 71
4.11 Rotor speed deviation response for mB & δB . . . . . . . . . . . . . . 71
4.12 Switching signal responses mE & δE . . . . . . . . . . . . . . . . . . . 72
4.13 Switching signal responses for mB & δB . . . . . . . . . . . . . . . . . 72
4.14 ∆δ and ∆ω responses for preliminary analysis . . . . . . . . . . . . . 77
4.15 Rotor angle & Rotor speed deviation respones . . . . . . . . . . . . 81
4.16 Switching signal response . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.1 Block Diagram of the optimal switching LQR compensator . . . . 90
5.2 Optimal switching LQG based Control . . . . . . . . . . . . . . . . . 92
5.3 ∆ω response of case I.(i) (a) & (b) . . . . . . . . . . . . . . . . . . . . 98
5.4 ∆ω response of case I.(i) (c) & (d) . . . . . . . . . . . . . . . . . . . . 98
5.5 ∆ω response of case I.(ii) (a) & (b) . . . . . . . . . . . . . . . . . . . 98
5.6 ∆ω response of case I.(ii) (c) & (d) . . . . . . . . . . . . . . . . . . . 99
5.7 ∆ω response of case II.(i) (a) & (b) . . . . . . . . . . . . . . . . . . . 99
5.8 ∆ω response of case II.(i) (c) & (d) . . . . . . . . . . . . . . . . . . . 99
vii
5.9 ∆ω response of case II.(ii) (a) & (b) . . . . . . . . . . . . . . . . . . . 100
5.10 ∆ω response of case II.(ii) (c) . . . . . . . . . . . . . . . . . . . . . . . 100
5.11 Switching signal response of case I.(i) (a) & (b) . . . . . . . . . . . 100
5.12 Switching signal response of case I.(i) (c) & (d) . . . . . . . . . . . 101
5.13 Switching signal response of case I.(ii) (a) & (b) . . . . . . . . . . . 101
5.14 Switching signal response of case I.(ii) (c) & (d) . . . . . . . . . . . 101
6.1 Block diagram of proposed switching system . . . . . . . . . . . . . 113
6.2 ∆δ & ∆ω responses for Light Load . . . . . . . . . . . . . . . . . . . . 122
6.3 ∆δ & ∆ω for Normal Load . . . . . . . . . . . . . . . . . . . . . . . . . 123
6.4 ∆δ & ∆ω responses for Heavy Load . . . . . . . . . . . . . . . . . . . 123
6.5 ∆δ & ∆ω responses for scenario 1 . . . . . . . . . . . . . . . . . . . . 123
6.8 Switching signal responses for ∆δ for scenario 1 . . . . . . . . . . . 124
6.10 Switching signal responses for ∆ω for scenario 1 . . . . . . . . . . . 125
viii
7.1 ∆δ & ∆ω responses for light load . . . . . . . . . . . . . . . . . . . . . 140
7.2 ∆δ & ∆ω responses for normal load . . . . . . . . . . . . . . . . . . . 141
7.3 ∆δ & ∆ω responses for heavy load . . . . . . . . . . . . . . . . . . . . 141
7.4 ∆δ & ∆ω responses for light load . . . . . . . . . . . . . . . . . . . . . 143
7.5 ∆δ & ∆ω responses for normal load . . . . . . . . . . . . . . . . . . . 143
7.6 ∆δ & ∆ω responses for heavy load . . . . . . . . . . . . . . . . . . . . 144
7.7 General switched linear system . . . . . . . . . . . . . . . . . . . . . . 145
7.8 ∆δ & ∆ω responses for case I . . . . . . . . . . . . . . . . . . . . . . . 148
7.9 ∆δ & ∆ω responses for case II . . . . . . . . . . . . . . . . . . . . . . . 151
7.10 ∆δ & ∆ω responses for case III . . . . . . . . . . . . . . . . . . . . . . 151
7.11 switching signal responses for case I . . . . . . . . . . . . . . . . . . . 152
7.12 switching signal responses for case II . . . . . . . . . . . . . . . . . . 152
7.13 switching signal responses for case III . . . . . . . . . . . . . . . . . . 153
ix
LIST OF TABLES
2.1 FACTS devices cost comparison . . . . . . . . . . . . . . . . . . . . . 32
2.2 Comparison of FACTS devies for stability . . . . . . . . . . . . . . . 32
4.1 Comparison of Peak Overshoots (Mp ) & Settling Time (Ts ) for
switching control algorithms . . . . . . . . . . . . . . . . . . . . . . . . 64
4.2 Comparison of Peak Overshoots (Mp ) for the UPFC control inputs. 72
4.3 Comparison of Settling Time (Ts ) for the UPFC control inputs. . 73
R∞
4.4 Comparison of Performance Index (J = 0
y 2 dt) for the UPFC
control inputs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.5 Comparison of performance index J with Other optimization tech-

niques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.6 Comparison of MP for PSS, UPFC and PSS+UPFC control inputs. 78
4.7 Comparison of TS for PSS, UPFC and PSS+UPFC control inputs. 78
4.8 Comparison of MP , TS & J for ∆δ. . . . . . . . . . . . . . . . . . . . 82
4.9 Comparison of MP , TS & J for ∆ω. . . . . . . . . . . . . . . . . . . . 83
5.1 UPFC control inputs for switching . . . . . . . . . . . . . . . . . . . . 89
5.2 Coordinated PSS & UPFC control inputs for switching . . . . . . 91
5.3 Modified controller gains for switching . . . . . . . . . . . . . . . . . 91
5.4 Comparison of peak overshoots, settling time & performance in-

dex J for Case I.(i) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
x
5.5 Comparison of peak overshoots, settling time & performance in-
dex J for Case I.(ii) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.6 Comparison of peak overshoots, settling time & steady state error
for Case II.(i) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.7 Comparison of Performance Index J for the Case II.(ii) . . . . . . 104
6.1 Operating Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
6.2 Modified Controller Gains . . . . . . . . . . . . . . . . . . . . . . . . . 114
6.3 Allowable System Matrix . . . . . . . . . . . . . . . . . . . . . . . . . 115
6.4 Control Input mB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
6.5 Control Input δB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
6.6 Control Input mE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
6.7 Control Input δE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
6.8 Peak Overshoots comparison for ∆δ . . . . . . . . . . . . . . . . . . . 129
6.9 Settling Time in seconds comparison for ∆δ . . . . . . . . . . . . . . 129
6.10 Peak Overshoots comparison for ∆ω . . . . . . . . . . . . . . . . . . . 130
6.11 Settling Time in seconds comparison for ∆ω . . . . . . . . . . . . . . 130
6.12 Performance Index J comparison for ∆δ . . . . . . . . . . . . . . . . 131
6.13 Performance Index J comparison for ∆ω . . . . . . . . . . . . . . . . 131
7.1 Comparison of Peak Overshoots (Mp ) & Settling time (Ts ) in sec-
onds for ∆δ and ∆ω with different LQR approaches for light load. 142
onds for ∆δ and ∆ω with different LQR approaches for normal
load. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
xi
onds for ∆δ and ∆ω with different LQR approaches for heavy load.142
7.4 Optimized feedback controllers for different loads . . . . . . . . . . 144
onds for ∆δ for case I . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
onds for ∆ω for case I . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
onds for ∆δ for case II . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
onds for ∆ω for case II . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
onds for ∆δ for case III . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
onds for ∆ω for case III . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
R∞
7.11 Comparison of performance index J = 0
y 2 dt for case I . . . . . . 150
R∞
y 2 dt for case II . . . . . . 153
R∞
y 2 dt for case III . . . . . 154
xii
Abbreviations
FACTS: Flexible AC Transmission System.
PSS: Power System Stabilizer.
LQR: Linear Quadratic Regulator.
SMIB: Single Machine Infinite BUS.
EPRI: Electric Power Research Institute.
SVC: Static Var Compensator.
TCSC: Thyristor Controlled Switched Series Capacitor.
TCPS: Thyristor Controlled Phase Shifter.
PSO: Particle Swarm Optimization.
UPFC: Unified Power Flow Controller.
PID: Proportional Integral Derivative.
ANN: Artificial Neural Network.
LFO: Low Frequency Oscillations.
GA: Genetic Algorithm.
IPSO: Iterative Particle Swarm Optimization.
STATCOM: Static Synchronous Compensator.
SSSC: Static Synchronous Series Capacitor.
SMC: Sliding Mode Control.
CPSO: Conventional Particle Swarm Optimization.
LQG: Linear Quadratic Gaussian.
LTI: Linear Time Invariant.
xiii
STR: Self Tuning Regulator.
IEEE: Institute of Electrical & Electronics Engineers.
MIMO: Multi Input Multi Output.
SISO: Single Input Single Output.
AVR: Automatic Voltage Regulator.
RCGA: Real Coded Genetic Algorithm.
ICA: Imperialistic Computative Algorithm.
BT: Boosting Transformer.
ET: Excitation Transformer.
VSC: Voltage Source Converter.
xiv
Notations
A: System or Evaluation matrix.
B: Control input matrix.
C: Observation Matrix.
D: Direct transmission matrix.
y: Output matrix.
K: Controller gain.
x: State variable.
u: Control law.
∆δ: Rotor angle deviation in degree.
∆ω: Rotor speed deviation in p u.

0
∆Eq : Quadrature-Axis component deviation in p u.
∆Ef d : Field voltage deviation in p u.
H∞ : H-infinity optimal control.
r: Reference input.
R & Q: Weighting matrices.
F1 & F2 : Switching boundary vectors.
∆: Deviation.
∆Tm : Deviation in mechanical torque.
KD : Damping constant gain.
W : Process noise matrix.
V : Measurement noise matrix.
xv
k1 − k6 : K constants of the linearized expressions.
H: Generator inertia constant.
KA : Gain of the excitation system.
TA : Time constant of the excitation system.
M : Machine inertia constant.

0
Tdo : D-axis transient open circuit field time constant.
ω0 : Synchronous speed.
∆mE : Deviation in modulating index of series inverter.
∆mB : Deviation in modulating index of shunt inverter.
∆δE : Deviation in phase angle of series inverter.
∆δB : Deviation in phase angle of shunt inverter.
J: Performance index.
MP : Peak overshoots.
TS : Settling time.
Vdc : DC link capacitor voltage.
AL : Light load system matrix.
BL : Light load control input matrix.
AN : Normal load system matrix.
BN : Normal load control input matrix.
AH : Heavy load system matrix.
BH : Heavy load control input matrix.
σ(t): Switching signal.
xvi
BP SS : Power system stabilizer control input matrix.
BU P F C : Unified Power Flow Controller (UPFC) control input matrix.
L: Kalman filter optimal gain.
Vt : Terminal voltage.
xvii
Optimal Switching Strategy for Power System with FACTS Based Controllers
Chapter 1
INTRODUCTION
ECE Department, SJCE, Mysore Page 1

1 INTRODUCTION
Optimal Switching Strategy for power system with Flexible AC Transmission

(FACTS) based controllers is an area of investigation for the electrical engineers from the
past few decades which concentrate mainly in the fields of designing optimal controllers,
switched systems and mathematical modeling of power system.
The mathematical optimization method for deriving control policies to an exten-

sion of the calculus of variations is referred to as optimal control theory. The technique
exists from Lev Pontryagin and his collaborators in the Soviet Union and Richard Bellman
in the United States. Optimal control take measures for searching a control law for a given
problem such that some required optimal criterion is achieved for that particular system.
Due to the combination of both continuous and discrete dynamics presently, there
is an abundant interest in hybrid or switching systems [1]. The study of switched linear
control theory is an area of research which especially concentrates on the systems and
control review from the past five decades. Switched linear control systems consists of set
of linear subsystems mentored by a switching control decision that take care of handling
this subsystems to change among them. Use of brilliant switching techniques in the area
of control was argument to give best performance when compared to the performance of a
system which does not use any switching. The switching control law operates by switching
between two closed loop controller gains in order to achieve the stability of that system
guaranteed as well as improvement performance in some metrics. The benefits of switching
between various subsystems are to utilize the needed properties of each subsystem and to
introduce new property which is not present in any one of the subsystem used.
For instance, there is mounting interest in optimal switching strategy, where the
control inputs u1 & u2 are designed with optimal control law by achieving some optimal
criterion and switching concept is introduced to switch between two individual optimal
controllers (u1 & u2 ) to achieve additional optimization by minimizing the output energy.
The generation of electrical energy, controlling of this energy for transmitting long

distances, utilization of this energy will be together called as power system. Today’s
power system is heavily loaded consisting of many networks connected with various buses
and generators. The review of earlier power transmission techniques are as follows:
• Existing power generation networks are usually not located near a load centre.
• To meet the rapidly increasing power requirements, beneficiaries are interested to use
maximum fully available power, instead of building a new constructing transmission
lines and enhancing substations.
• The transmission lines will overload due to power flows in that lines, which has as
an overall drawback of changing voltage profiles and chances of system instability.
• In most cases, the available transmission networks will fail to handle the control
requirements of today’s complex and heavily loaded power networks.
This above mentioned overall situation motivated to propose new transmission

control approaches, which has been carried out in this thesis.
With the increase in electric power system networks, power system oscillations are
also increases with few cycles per minute. These oscillations are mainly due to the lack of
damping of mechanical mode of the system. If these oscillations are not controlled, they
will grow enormously in magnitude until loss of synchronize results [2]. In order to damp
these developed oscillations and to increase system stability, incorporating power system
stabilizer (PSS) is both inexpensive and effective. PSSs have been used for many years to
control the developed oscillations in power system. But, the PSS fails to control the great
variations in the voltage profile and can even result in increasing of power factor and loss
of system stability under severe disturbances.
Recently, with the rapid improvements in power electronics has opened the new
door for the application of Flexible AC Transmission Systems (FACTS) for power system
networks. FACTS are a combination of various power electronics device, which can be

used individually or coordination with one another to control one or more power system
inter related parameters [3].
The complexity of interconnection of switching strategy with power system de-

vices such as FACTS poses new challenges in modeling and analysis for the improvements
in overall performance of the power system. Previous attempts to model such systems
(without switching) have relied on tools from Linear Quadratic Regulator (LQR) theory
[4], Linear Matrix Inequality [5], H∞ optimization theory [5] and Genetic Algorithm (GA)
[6] etc.
The aim of this research is to develop a hybrid modeling for the synchronous ma-
chine infinite bus (SMIB) power system with the switching operation of FACTS devices,
using various combinations of novel switching techniques. This hybrid model is tested for
improved performance under switching conditions over the nominal operation by simulat-
ing on MATLAB/SIMULINK
R platform.
1.1 Literature Survey
In this subsection, literature survey of power system with FACTS based controllers,
switched linear systems and switching control strategies for power system is given.
1.1.1 Power System with FACTS Based Controllers
The FACTS concept was developed by Electric Power Research Institute (EPRI) in the
year 1980. N. G. Hingorani et. al [7] introduced this new concept for the existing resources
of power systems, initiating a new direction in power system research.
The Thyristor has been applied for high voltage power systems due to its high
voltage & current ratings have been demonstrated in [8-9]. Wang and Swift et al. [10] de-
veloped a novel unified Phillips-Heffron model for a power system installed with Static Var
Compensators ( SVCs), Thyristor Controlled Switched Series Capacitors (TCSCs/TSSCs),

and Thyristor Controlled Phase Shifter Regulators (TCPSs). Abido & Abdel-Magid [11-
12] shows the improvements with PSS and FACTS based controllers for the application
of power system. In the area of designing controllers for FACTS devices, Majority ap-
proaches are based on the design of advanced control theory for the application of TCSC
controller [13-22].
S. Eshtehardiha & Gh. Shahgholian et al. [23] presented improved performance

in STATCOM output current by designing the controllers using two methods LQR & Pole
placement. The experimental results demonstrated that the response of currents control
are improved with simple LQR & pole placement methods. Akram in [24], proposed the
coordinated design of PSS & FACTS device and shows the improvements in damping
compared to uncoordinated PSS & FACTS device.
R. K. Pandey et al. [25] proposed multi-stage LQR concept for the multi machine
systems. The experimental simulation results shows that settling time is improved with
minimum overshoots. In [26], the authors proposed a novel method of designing a power
system oscillation damping controller for UPFC. The design problem of this work is based
on θ - PSO optimization control with choosing eigenvalues has objective function.
Sasongko Pramono Hadi et al. [27] provides a detailed dynamics for modeling of
a multi machine power system installed with Genetically based UPFC as the extension
of UPFC configuration. Based on this model, the authors demonstrated the validity
of this dynamic model by numerical simulations. In [28], the authors proposed a three
robust control methods Quantitative Feedback Theory (QFT), µ-synthesis and H∞. The
proposed three control methods are applied to a SMIB based UPFC in various loading
conditions. The simulation results demonstrated that for all the loading conditions the
proposed controllers will provide robust performance.
Doradla & Prathap Hari Krishna et al. [29] proposed three different types of
PSS using modified Phillips heffron model. The authors demonstrated that the proposed
method of three PSS are well suited for the application of power system at different

system operating conditions. Cuk Supriyadi & Ali Nandar et al. [30] proposed the design
of robust PSS considering less control energy into account. The authors demonstrates the
performance of controller was guaranteed for the power system oscillation damping with
less control energy.
A Venkateshwara Reddy et al. [31] proposed a new optimal LQR PSS. The pro-
posed controller demonstrates that it can be realized by the local measurements and does
not require any external system information. Balwinder Singh Surjan et al. [32] con-
sider the results obtained for different possible combinations of the controllers and finally,
concluded that PID controller in combination with other controller is effective for the
improvement of settling time and Integral Square Error (ISE).
Sangu Ravindra et al. [33] proposed a adaptive Artificial Neural Network (ANN)
damping controller for SMIB based UPFC. The simulation results of the proposed control
technique are compared with simulation results of conventional lead-lag controller.
A.V. Sudhakara Reddy et al. [34], designed an adaptive neuro-fuzzy controller

for UPFC with effect to three different loading conditions. Comparison results shows
that the proposed controller has improved settling time and peak overshoots compared to
conventional lead lag controller. The authors in [35] investigated low frequency oscillation
damping using a UPFC. The proposed UPFC control design problem was converted into
an optimization problem and solved using Gravitational Search Algorithm (GSA) and
concludes that the UPFC control input δE provides best performance in damping compared
to other control inputs of UPFC.
Balwinder Singh Surjan and Ruchira Garg et al. [36] have proposed the effective-
ness of conventional PSS and PID-PSS for the SMIB based PSS and comparison results
shows that the PID-PSS provides better performance compared to conventional PSS. In
[37], the authors proposed particle swarm optimization based technique for the develop-
ment of tuning the parameters in fixed structure PSS. The proposed algorithm concludes
that the designer has a flexibility to achieve a trade off between the overshoots and control

constraint. Ali M. Yousef and M K Ei-Sherbiny et al. [38] have designed LQR based PSS.
The proposed LQR-PSS has robustness control property with power system prominant
parameters. The IPSO algorithm was introduced in [39]. This proposed IPSO was uti-
lized to find the optimal parameters of PSS for SMIB system by minimizing the objective
function. Using the proposed algorithm, the authors proved that LFO can be reduced
appropriately.
In [40] the authors, presented a free model approach for system identification and
its application to design a PSS. The design of optimal controllers have been proposed by
Ali. M. Yousef and Ahmed M Khan in [41] for PSS to enhance the synchronizing and
damping torque coefficients.
Sai Shankar, K T Veeramanju & Yathisha L et al. [42] proposed different opti-
mal LQR controllers by tuning the weighting matrices Q & R for the improvement of
STATCOM performance. Ali. M. Yousef and Mohamed Zahran et al. [43] have designed
a Linear Quadratic Gaussian (LQG) optimal control for PSS and showed the improved
damping compared to conventional LQR optimal control.
The Unified Power Flow Controller (UPFC) has more than one function FACTS
device, whose master duty is to control the flow of power in power system networks. The
alternate duties of UPFC are voltage control, stability improvements in transient periods,
damping of oscillations etc. It combines the properties of series and shunt type FACTS
devices.
H.F. wang et al. [44] presented a linearized Phillips Heffron Model installed with
UPFC for single machine infinite bus (SMIB) in power system. Presently, researchers are
working for choosing the best control inputs from UPFC such as mE , δE , mB & δB to
cover wide range of operating conditions by applying various novel control techniques.
some of the examples are described here. H. Shayeghia & H.A. Shayanfar et al. [45] has
designed output feedback control technique for UPFC using particle swarm optimization
and concludes that UPFC control input δE was the most robust control compared to other

UPFC control inputs.
The authors in [46] proposed phase compensation control technique for the four
UPFC control inputs and reveals that the UPFC control inputs δE & δB has improved
performance compared to the other three UPFC control inputs. A.K. Baliarsingh & S.
Panda [47] have applied a real-coded genetic algorithm optimal control concept for UPFC
and investigation of this work concludes that damping control mB provides little better
performance among four alternatives of UPFC control inputs.
Vitthal Bandal and B. Bandyopadhyay et al. [48] proposed, the design of PSS
for SMIB power system based on fuzzy logic and output feedback sliding mode controller
(SMC). It concludes that the designed controller provides improved damping of power
system oscillations.
The application of the UPFC to the advanced power system can significantly in-
crease flexible operation, secure and economic [49-51]. In [52], the authors presented itera-
tive particle swarm optimization (IPSO) method based on UPFC controller to achieve im-
proved robust performance compared to conventional particle swarm optimization (CPSO)
approach.
The review of earlier research reveals that, uncoordinated control design of PSS
& FACTS devices may cause destabilizing interactions. To overcome this drawback and
to improve system performance many researches are working on the coordinated design of
PSSs and FACTS damping controllers [53-54] .
Kwang M. Son and Jong K. Park et al. [55] have applied the LQG based opti-
mal control technique for the design of TCSC controller to damp & enhance the power
system oscillations. UPFC auxiliary stabilizer by using LQG control strategy is proposed
by Amir Elahi and Alireza Gholizadeh in [56]. The effectiveness and validity of the pro-
posed method was evaluated with various different simulations. The digital results show
the improvements in damping of power system oscillations under system is subjected to
disturbances.

1.1.2 Switched Linear Systems
Switched linear systems is a class of hybrid systems, we have used as tool for modeling in
this thesis. Different authors considered the concept of switched linear systems in various
ways. For example Daniel Liberzon [57], [58] used differential geometric approach based on
common quadratic lyapunov function to find sufficient conditions for asymptotic stability
of linear time invariant system.
The authors in [59] reviews the advanced developments in the analysis and syn-
thesis of switched linear control systems and demonstrates the qualitative properties of
switched linear control systems such as controllability, observability and design of optimal
controllers with reference to switched systems. The principle of duality in switched linear
system was proposed by Zedong Sun in [60]. In yet another paper by Zedong Sun et
al. [61] demonstrated that the controllable & observably are synthesized as in the same
manner of Linear Time Invariant (LTI) systems.
A. Nerode et al. [62] proposed the logic based driven sliding mode control as hybrid
control that is the mixture of continuous and discrete variables. A. Ferrate, L. Magnani
and R. Scattlolini proposed a hybrid variable structure control strategy in their work [63].
But, the strategies applied in this work are sliding modes for the variable structure control.
Figure 1.1: Multicontroller Architecture of Switched Systems

The multi controller architecture for switched linear systems is as shown in Figure
1.1, where the switching takes place between different controllers as per the instruction of
supervisor to the process.
M. Barnicky et. al [64] presents the stability analysis with finite switches in finite
time for multi-modal systems. Liberzon and Morse et al. [65] described the basic idea for
designing switched linear systems. The authors in [66-68] demonstrates some recent works
with switched linear systems using polyhedral/polyhedral Lyapunov functions.
Zhi Hong Huang & Cheng Xiang et al. [69] derived a necessary and sufficient con-
dition for stability of arbitrarily switched second order LTI systems with marginally stable
subsystems. It turns out that the condition for the marginally stable case is similar with
the one for asymptotically stable except boundary conditions are included. The perfor-
mance of switched linear systems can be realized by designing a control law for the purpose
of switching between two LTI subsystems [70-72]. The performance based switching con-
trol law for two LTI subsystems based on Lyapunov stability criteria was developed by
Aravena & Devarakonda in [73]. Keith R Santarelli et al. [74] developed a state feedback
switching control law for the stabilization of two LTI subsystems using pole placement
method. In [75], the authors derived the switching control law for the minimization of
performance index to achieve some optimization by minimizing the output energy when
switching takes place between two LTI systems compared to two individual LTI systems.
In [76], the authors designed an concept of arbitrary switching with constructing the cost
function and solved using genetic algorithm.
A brief literature survey of more recent works in optimization via switching control
is also discussed to motivate the proposed research.
Tuhin Das & Ranjan Mukherjee et al. [77] address the optimal switching problem
for switched linear systems and derived the condition for optimal switching by embedding
the switched system in a larger family system by applying pontryagins maximum principle.
Keith R Santarelli & Munther A Dahleh et. al [78] derived an switching algorithm which

optimizes the rate of convergence by employing a controller that switches between two
linear subsystems, one of which is unstable.
Shengxiang Jiang & Petros G. Voulgaris et. al [79] demonstrated that optimization
is achieved for Youla-Kucera parameters, where switching occurs as usual with respect
to time. In [80], the authors designed optimal switching control for voltage converter
circuits, with minimizing the cost function as switching energy. Simulation results exhibit
fast convergence of the algorithm, and suggest its potential utility in a broader class of
switching-control applications in power electronics.
1.1.3 Switching Control Strategies for Power System
Hussain N and Al-Duwaish et al. [81] have designed PSS with switching control using
adaptive neural network Sliding Mode Control (SMC) and simulation results concludes
that performance is improved with this switching control technique. In [82] the authors
demonstrate the switching control for PSS, which has adopted less energy for designing
controller. In [83], for the purpose of multi-identification models the authors proposed a
self-tuning regulator (STR) with minimum variance using fuzzy logic switching. Vittal
Bandal and B Bandyopadhyay et al. [84] have proposed the design of PSS with two
switching methods based on fuzzy logic & output feedback SMC and investigation reveals
that the proposed switching controllers has well damping enhancement.
1.1.4 Drawbacks of the Existing Control Techniques for Power System
The earlier conventional control techniques like phase compensation, pole placement, PI,
PD & PID controllers applied by the researchers for the design of PSS & FACTS devices
in power system simply makes a system without optimizing anything. These conventional
control techniques are also designed based on the transfer function approach not state
space approach i.e., we can see only the input & output relation, but we cannot see what
is happening inside the system to improve the overall performance of the power system.

Later, the researchers are started to design their new control techniques using ad-
vanced state space approach for the power system model installed with PSS & FACTS
devices. The new control techniques designed by the researchers are based on the optimiza-
tion techniques like LQR, LQG, Genetic Algorithm, Tabu Search Algorithm (TSO), Par-
ticle Swarm Optimization (PSO), Iterative PSO, H∞ , Artificial Neural Networks (ANN),
Fuzzy Logic etc,. These optimal control techniques have advantageous that it will makes
the system stable as well as some optimization is achieved based on the setter performance
index J is minimized. The drawbacks of these existing optimization control techniques to
propose switching control techniques for power system are:
1. To, achieve better performance in overshoots & settling time by intelligently switch-
ing between two optimal controllers.
2. UPFC has four control inputs (mB , δB , mE , δE ), the researchers are selecting better
UPFC control inputs by applying their optimal control techniques. The four UPFC
control inputs have their own advantageous compared to other UPFC control inputs.
Instead of selecting one UPFC control input, switching between two UPFC control
inputs is proposed to combine the advantages (properties) of both UPFC control
inputs.
3. Due to huge loading conditions power system always operates at different operating
conditions. In this case, selecting or designing single optimal control for all the
operating conditions is near impossible. For, this problem multi optimal controllers,
are designed and each optimized at different operating conditions and to develop a
intelligent switching technique to select appropriate optimal controller into feedback
at different operating conditions.
Recently, switching control strategies are developed for power system. From, the
literature review it reveals that only very few works are done with respect to designing
switching control strategies for the applications of power system. The switching control

techniques applied by the researchers are sliding mode control (SMC) & fuzzy logic switch-
ing for PSS only not for FACTS devices in power system. These sliding mode control &
fuzzy switching occurs at very high rate leading to chattering effect. Hence, it is not
suitable for optimal control applications, particularly in power systems.
1.2 Objectives of the Research
The main objectives of this research consists of five parts:
1. To model power system installed with FACTS devices such as UPFC.
2. Apply Switching Algorithms to power system with PSS & UPFC and achieve im-
provements in performance parameters such as overshoots, settling time and output
energy compared to systems without any switching.
3. Apply Switching Algorithm to power system with UPFC with controllers designed
using LQR & LQG.
4. To model power system with UPFC operating at different loading conditions.
5. Apply Switching Algorithm to power system with UPFC operating at different load-
ing conditions and to achieve improvement in performance compared to systems
without switching.
1.3 Problem Statement
The usage of available power system networks can be improved with the application of
modern power electronics technologies. Flexible AC Transmission Systems (FACTS) pro-
vide technical solutions to address the new operating challenges being presented today.
The potential benefits of FACTS equipment are nowadays widely recognized by

the power systems engineering community, however, the current challenge is how to obtain

the maximum benefit from these devices such that a certain optimal criterion is achieved.
This optimization problem includes these aspects: finding the optimal control law
for the power system model and optimizing its state variable parameters such that the
maximum benefit can be obtained to improve the overall performance of the system, in
terms of minimizing output energy, overshoots and settling time.
1.4 Scope, Limitations and Delimitation’s of the Research
The proposed research has very broad scope for research activity as there are very few
reported works done on switched systems, particularly employing a switching strategy in
the context of applications to flexible AC power transmission systems. The product of
this research, both the power systems with FACTS controllers and the switching strategy
for power system with FACTS controllers will contribute towards the overall performance
of the system.
Proposed research will further aid in understanding the issues of switched sys-
tems in various large scale flexible AC transmission systems operating in uncertain envi-
ronments, which may require more sophisticated modeling involving stochastic or fuzzy.
Hence the proposed research will be helpful in theoretical and practical knowledge in the
areas of control systems and power systems.
The delimitations set for the proposed research is to apply the switching control
techniques for the linearized power system models instead of non-linear algebraic equations
of power system, which requires more analysis. For, the small signal (steady state) stability
analysis linearized differential equations are sufficient. To, know the stability analysis of
large disturbances (transient stability) non linear algebraic equations are must needed
which can be solved using some iteration techniques. The limitations of current research
are there is no rigorous mathematical proof of optimization for switching control algorithm
I, used in this thesis. Analysis is extendible to multi machine network scenario, but has
not been explored in this thesis.

1.5 Thesis Outline
This thesis is organized as follows: Chapter 2, Introduction to power system stability,

Linearization concepts for power system, Comparison review of various FACTS devices
is described. Finally, the basic design procedures of Phillips heffron model starting from
the swing equation to state space model of Phillips heffron model for power system are
explained in detailed.
In chapter 3, Optimal control theory of LQR & LQG control for LTI systems are
discussed followed by basic overview of switched linear time invariant (LTI) systems with
motivation for switching. The Intelligent switching control algorithms which are applied for
the current research with suitable modifications based on the switching requirements are
briefed along with the stability proof of these switching algorithms. The basic knowledge
of modeling of power system in chapter 2, switched linear systems and optimal control
theory design of present chapter are very essential for developing the switching strategies
with optimal controllers design for the power system installed with PSS & FACTS devices
in the subsequent chapters.
Main contributions of the research are discussed extensively from Chapter 4 to

Chapter 7.
Chapter 4, describes the linearized state space model (Phillips Heffron SMIB
model) installed with PSS & UPFC . After, we focus on application of two switching
control algorithms for the PSS. Switching between two feedback controllers (optimized &
non-optimized feedback controllers) is suggested for the Phillips heffron model installed
with PSS. The simulation results obtained from the two modified switching control algo-
rithms are compared with respect to the prominent parameters of power system such as
rotor angle & rotor speed deviations. Later, the best modified switching control algorithm
from the comparison is applied for the Phillips heffron SMIB model installed with UPFC
to switch between the two optimized feedback controllers for all the four UPFC control in-
puts (mB , δB , mE & δE ) and the results are compared with individual optimal controllers

as well as other optimized controllers for the UPFC in power system. Implementation of
switching control techniques for the coordinated design of PSS & UPFC is described in
the later sections of this chapter. Initially, a preliminary optimal control analysis is done
between PSS & UPFC to show the effectiveness of coordinated design of PSS & UPFC
compared to uncoordinated design of PSS & UPFC. Later, by considering the concepts &
future direction of the works in [74,94], a new switching technique are developed to switch
between optimized & non-optimized feedback controllers for the coordinated design of PSS
& UPFC. The simulation results are compared with the individual feedback controllers.
The results in this chapter are based on the dissemination work paper 2, 3 & 4.
In Chapter 5, the optimal switching strategies for various combinations of UPFC

control inputs (mB , δB , mE & δE ) with four cases are developed. In the first case switching
between two uncoordinated UPFC control inputs is proposed in various combinations.
Second case is developing the switching control techniques for various combinations of
UPFC control inputs coordinated with PSS. Case 3 experiments is conducted to show the
effectiveness of LQG optimal control compared to LQR based optimal when the power
system with UPFC is subjected to disturbances. Finally, fourth case is experimented for
the development of switching control between two LQG controllers when the power system
with UPFC subjected to disturbance. The results of all the four cases are individually
compared as well as overall and drawn to relevant conclusion. The results in this chapter
are based on the dissemination work paper 7.
Chapter 6, concentrates on developing optimal switching techniques for three oper-

ating conditions (light, normal & heavy loads) of power system. Designing a single optimal
control for today’s complex power system is not always optimal for all the operating con-
ditions. To overcome this drawback, in this chapter individual controllers are optimized
for all the specified operating conditions and switching technique is introduced to switch
between two operating conditions based on three scenarios to consider the appropriate
optimized controllers for that operating point to achieve overall robust performance for
all the operating conditions. The results of three obtained scenarios are compared with

individual loads for the selection of best UPFC control input among the four UPFC con-
trol inputs (mB , δB , mE & δE ).The results in this chapter are based on the dissemination
work paper 1.
In Chapter 7, Multi-stage switching approach is proposed for multi-LQRs of UPFC

based power system. This chapter apply the Bryson, Bouderal & multistage-based Linear
Quadratic Regulator (LQR) optimal controllers to the UPFC for light, normal & heavy
loads to cover various operating conditions as a preliminary analysis. The two best opti-
mized feedback controllers are selected from the preliminary analysis and optimal switching
strategy is implemented to further optimize of output energy for the specified operating
conditions. The current challenge in the present chapter is to design multi controllers,
each optimized at different operating conditions and to develop a switching algorithm to
select appropriate controller into feedback at different operating conditions.
Finally, chapter 8, discusses the conclusions and recommendations for further work
to carry in future. The overall summary of research carried throughout the thesis for the
present study are concluded as contributions. It explains the effectiveness of switching
control theory for the application of power system to damp the power system oscillations.
Further, some suggestions on the extension of the proposed work for future research are
proposed.
1.6 Summary
This chapter explains the introduction to theories of switched linear, optimal control and
power system. A brief literature survey is carried to get the basic knowledge on both
power system as well as switched linear control systems. In order to motivate the proposed
work the drawbacks of existing control techniques for the application of power system are
detailed. Objectives, scope, limitations, delimitation’s and problem statement for the
proposed research are also discussed in this chapter. Finally, thesis organization from the
next preceding chapters is detailed to get an overview of each chapter.

Chapter 2
PRELIMINARIES

2 PRELIMINARIES
2.1 Introduction
In order to implement switching strategies for power system, it is essential to gain basic
fundamentals of both power system as well as FACTS devices.
This chapter explains the basic power system non linear model with an overview
of stability analysis. Linearization concepts are also described in this part of the chapter.
Introduction to FACTS controllers, types along with comparison of FACTS devices is also
described. Finally, Phillips heffron model design is explained in this chapter.
2.2 Power System
Power systems are referred to as real time energy systems which generate, transmit and
supplies power the moment you turn on the light switch. The power system dynamics are
modeled using the combination of both nonlinear differential and algebraic equations.
The power system Differential Algebraic Equations (DAE) consists of both sets of
equation. The DAE model as as follows:
ẋ = f (x, u, N ) (2.1)
0 = g(x, u, N ) (2.2)
p = h(x, N ) (2.3)
Where,
f = Group of non linear equations which is dependent on model.
g = Group of non linear algebraic equations.
x(t) = Group of state variables.
u(t) = Group of control inputs.

h = Group of nonlinear algebraic equations which relates the measurement of

states.
p = Group of system measurements.
N = Group of Network Parameters.
2.2.1 Linearization Concepts
Power system dynamic behavior is described by a group of n first order nonlinear ordinary
differential equations with the following form: n= order of the system; r= number of
inputs.
ẋi = fi (x1 , x2 , ....., xn ; u1 , u2 , ....., ur ; t) i = 1, 2, ....., n (2.4)
ẋ = f (x, u, t) (2.5)
     
x1 u f1
   1   
 x2   u2  f2 
     

     
Where, x= .  u= . f = . 
     

     
     
 .   .   . 
     
xn ur fn
Where, x & u is referred to as state and control input vectors in column. f is
referred to function vector with respect to time. Eq. 2.6 shows the state variable x, which
does not depends on time t.
ẋ = f (x, u) (2.6)
The output of the system is given by,

y = g(x, u) (2.7)
Let, for the small variations Eq. 2.6 can be written as follows:
x = x0 + ∆x u = u0 + ∆u
where, ∆ represents a small deviation.
ẋ = x˙0 + ∆ẋ = f [(x0 + ∆x), (u0 + ∆u)] (2.8)
Let x0 & u0 be the initial state and input vectors. Considering the variations are
small, the Eq. 2.8 can be written as follow:
ẋi = x˙i0 + ∆ẋi
= fi [(x0 + ∆x), (u0 + ∆u)]

∂fi ∂fi ∂fi ∂fi
= fi (x0 , u0 ) + ∆x1 + ..... ∆xn + ∆u1 + ..... + ∆ur
∂x1 ∂xn ∂u1 ∂ur
Since ẋi0 = fi (x0 , u0 ), we obtain
∂fi ∂fi ∂fi ∂fi

∆ẋi = ∂x1
∆x1 + ..... ∂xn
∆xn + ∂u1
∆u1 + ..... + ∂ur
∆ur
with i = 1, 2, ....., n. In a like manner, from the Eq. 2.7, we have
∂gj ∂g ∂gj ∂gj

∆yi = ∂x1
∆x1 + ..... ∂xnj ∆xn + ∂u1
∆u1 + ..... + ∂ur
∆ur
with j = 1, 2, ....., m. The linearized forms of Equations 2.6 and 2.7 are
∆ẋ = A∆x + B∆u (2.9)
∆y = C∆x + D∆u (2.10)

where,
   
∂f1 ∂f1 ∂f1 ∂f1
∂x1
..... ∂xn ∂u1
..... ∂ur
   
   
   
A=
 ..... ..... .....  B=
 ..... ..... ..... 
 
   
 ∂fn ∂fn   ∂fn ∂fn 
∂x1
..... ∂xn ∂u1
..... ∂ur
   
∂g1 ∂g1 ∂g1 ∂g1
.....
∂x1 ∂xn
.....
∂u1 ∂ur
   
   
   
C= ..... ..... .....  D= ..... ..... ..... 
   
   
 ∂gm ∂gm   ∂gm ∂gm 
∂x1
..... ∂xn ∂u1
..... ∂ur
The above partial derivatives are evaluated at the equilibrium point about which
the small perturbation is being analyzed. In Equations 2.9 and 2.10,
∆x = state vector deviation of dimension n.
∆y = output vector deviation of dimension m.
∆u = input vector deviation of dimension r.
A = state matrix of size nXn.
B = control matrix of size nXr.
C = output matrix of size mXn.
D = direct transmission matrix which appears directly in the output, size mXr.
2.2.2 Power System Stability
Power system stability is defined as the ability of synchronous generator to maintain

synchronize even after the disturbance. The stability limit is the maximum power that
can be transferred in a power network between sources and loads without loss of synchro-
nization [85-87].

Loss of Synchronization:
Loss of Synchronization is due to the following reasons:
1. Any unbalance between the sending and receiving end (refer Figure 2.1) developed
by a change in load demand, in generation, or in a network conditions.
2. Any unbalance between the generation and load initiates the rotors of the syn-
chronous machines to swing because net accelerating (or decelerating) torques are
developed on these rotors.
For the purpose of analysis the power system stability problems, the stability
conditions may be classified (refer Figure 2.2) as follows:
1. The steady state stability is the ability of the power system to maintain synchro-
nization even after the system is interpreted by small disturbances.
2. The transient stability is the ability of the power system to maintain synchroniza-
tion even after the system is interpreted by large disturbances.
Figure 2.1: Power Network
In present power systems networks, the small-signal stability problem is more

compared to large signal stability problem. The small signal stability analysis is examined
by applying linear concepts for the dynamic characteristics of power system. Steady state
stability (Small signal stability) analysis is subdivided into static stability and dynamic
stability.

Figure 2.2: Overview of stability analysis
2.2.3 Phillips Heffron SMIB Model
Phillips Heffron model of SMIB is commonly used for small signal stability analysis in
power system. To analyze the damping effect of PSS & FACTS based power system, this
model has been successfully applied from the last few decades. In the theory of control
system, for the purpose of stability analysis with designing advanced controllers linearized
differential equations of the physical systems are modeled referred to as state space model
and is defined as:
ẋ(t) = Ax(t) + Bu(t) (2.11)
y(t) = Cx(t) + Du(t) (2.12)

Figure 2.3: General Linear Systems
The Phillips heffron has analyzed the dynamics of power system and modeled with
the max-wells linearized differential equations using state space approach called Phillips
heffron model. The Linearized generator swing equations are given by
1
∆ω = (∆Tm − ∆Te ) (2.13)
2Hs + KD
2πf0
∆δ = ∆ω (2.14)
s
The ∆δ & ∆ω represents deviations in rotor angle and rotor speed. Figure 2.4
shows the Automatic voltage regulator (AVR) connected to a synchronous generator.
Figure 2.4: AVR connected to Generator
The phasor diagram of single machine infinite bus (SMIB) with the direct (d-axis)
and quadrature (q-axis) is as shown in the Figure 2.5. To maintain the power system

stability under all the conditions the power angle δ shown in Figure 2.5 should always be
within 900 . If this power angle exceeds 900 , than the power system is said to be unstable.
Figure 2.5: Phasor diagram of SMIB systems
Figure 2.6 shows that damping constant KD will sense the disturbances appearing
across ∆ω and correct that distortions with the help of change in electrical torque (∆Te ).
Figure 2.6: Phillips heffron model (a)
Due to change in electrical torque (∆Te ) the deviation in rotor angle (∆δ) will
starts to vary if not control this angle it exceeds above 900 . For, this the K-constant K1

is used as feedback signal (Figure 2.7) to control this angle with the help of another K-
constant compensation K2 .
Figure 2.7: Phillips heffron model (b)
Approximation of torque with power is given by
Te = Pe = id ud + iq uq (2.15)
After linearization and some substitutions
0
∆Te = K1 ∆δ + K2 ∆eq (2.16)
0
where, ∆eq is the change in quadrature axis component deviation.
Field voltage equation is represented as
0 0 0 0
Tdo eq + eq + (xd − xd )id = −eF (2.17)

0 0
(1 + sTdo K3 )∆eq = −K3 (∆eF + K4 ∆δ) (2.18)
The feedback signal K4 shown in Figure 2.8 will sense the internal voltage varia-
tions which will directly affect the rotor angle and correct it with the help of field winding
compensation as shown in Figure 2.8.
Figure 2.8: Phillips heffron model (c)
Figure 2.9 shows the full Phillips heffron model with the additional feedback signals
with gains K5 & K6 . The feedback signal with gain K5 will get the knowledge of any
0
variations in ∆δ and that with gain K6 will sense the variations in ∆Eq and correct this
variations with the help of excitation systems, AVR, additional damping parameters etc.
Terminal voltage is given by
u2t = u2d + u2q (2.19)
0
∆ut = K5 ∆δ + K6 ∆eq (2.20)

Figure 2.9: Phillips heffron model (d)
2.2.4 Flexible AC Transmission Sytems (FACTS)
FACTS developed by EPRI in late 1980, is the collection of various power electronic de-
vices which can be applied individually or coordinated with one another to control power
flow in transmission network, to improve system performance by damping oscillations, etc.
Utilizing a supplementary feedback control, along with the FACTS-device as primary con-
trol, has been proved that the system damping can be improved. Presently various series
and shunt type FACTS devices have been introduced for various applications in world-
wide. The development of FACTS controllers has followed two distinct groups, the first
group employs Thyristor controlled and second group employs Converter based FACTS
controllers.
A. Thyristor controlled FACTS controllers
The first group of FACTS controllers involve Static VAR Compensator (SVC),
Thyristor- Controlled Series Capacitor (TCSC) and phase shifter, Each of these controllers
can act on one of the three parameters determining power transmission, voltage (SVC),
transmission impedance (TCSC) and transmission angle (phase shifter), as illustrated in
Figure 2.10.

Figure 2.10: Conventional Thyristor Based FACTS Controller
B. Converter based FACTS controllers
This FACTS controller group employs Static Synchronous Compensators (STAT-

COM), Static Synchronous Series Compensators (SSSC), Unified Power Flow Controllers
(UPFC), and Interline Power Flow Controllers (IPFC). Figure 2.11 shows the overview of
synchronous voltage source operated as a STATCOM and Figure 2.12, represent the block
diagram of UPFC with two back-to-back voltage source converters.
Figure 2.11: STATCOM voltage source

Figure 2.12: UPFC configuration
The interest measure of science & engineering community in FACTS is shown in

Figure 2.13. The literature survey carried out in [88-89] shows that the number of pub-
lications, applications of FACTS to power system. From Figure 2.13, it is clear that the
interest in the converter based FACTS controllers (2nd generation) has rapidly increased
while the interest in the Thyristor controlled FACTS controllers (1st generation) has de-
creased.The comparison of various FACTS devices with respect to cost and power system
stability are as shown in Tables 2.1 & 2.2 [88-89] and it is clear that UPFC is one of the
most promising devices in FACTS concept.
Figure 2.13: FACTS applications year wise chart

Table 2.1: FACTS devices cost comparison
Sl. No FACTS devices cost (Rs./k var)

1 Shunt Capacitor 434
2 Series Capacitor 1090
3 SVC 2165
4 TCSC 2165
5 STATCOM 2710
6 Series type UPFC 2710
7 Shunt type UPFC 2710
Table 2.2: Comparison of FACTS devies for stability
Sl. FACTS Power System Load Voltage Transient Dynamic

No Devices Stability Enhancement flow control stability stability
1 UPFC Yes Large Large Average Average
2 TCSC Yes Average Small Large Average
3 SVC Yes Small Large Small Average
4 SSSC Yes Small Large Average Average
5 STATCOM Yes Small Large Average Average

2.3 Summary
In this chapter introduction to power system along with the linearization concept are
discussed. Power system stability with classification is briefly explained. Phillips hef-
fron SMIB model is derived starting from the generator swing equation. Finally, FACTS
controllers are discussed along with the comparison various FACTS devices. From, the
comparison of FACTS devices it is clear that the UPFC based FACTS device provides
better performance in all the stages. Hence, the research is concentrated on UPFC based
FACTS device in next preceding chapters.

Chapter 3
OPTIMIZATION USING SWITCHING

CONTROL

3 OPTIMIZATION USING SWITCHING

CONTROL
3.1 Introduction
This chapter explains the optimal controller design, switched linear control theory, motiva-
tion for switching, switching control algorithms along with the stability proof are detailed.
Optimization concepts of switching control algorithms are also derived to develop a opti-
mal switching strategy for power system using FACTS based controllers.
3.2 Optimal Control Design
Optimal control theory is defined as searching a control law for the given optimization
problem which is in the form of quadratic to obtain some optimization solution for that
particular system. Advantage of using optimal control technique compared to conventional
pole placement technique is that it provides a coherent method of evaluating the state
feedback control gain matrix. The optimized feedback controllers for the present research
are derived from the LQR algorithm and when the system is subjected to disturbances,
the controllers are derived from LQG algorithm. For, the sake of completeness the LQR
& LQG control methods are explained briefly [90-93]:
3.2.1 Linear Quadratic Regulator Control
The LQR control selects the feedback control matrix such that it minimizes the perfor-
mance index J by achieving the trade off between the use of control effort and the response
of the system that will guarantee a stable system [42 & 91].

Consider a LTI system with
ẋ(t) = Ax(t) + Bu(t) (3.1)
y(t) = Cx(t) (3.2)
The input u is expressed as r − Kx, where r is the reference input and K is the
feedback gain, also called the control law. Now assume that the reference input r is zero
and that the response of the system is excited by nonzero initial state x(0), which in turn
excited by external disturbances. If r = 0, then the input u = −Kx and the closed loop
system is given by
ẋ(t) = (A − BK)x(t) (3.3)
The most systematic and popular method is to find K to minimize the quadratic
performance index given by,
Z ∞
J(x, u, Q, R) = (xT Qx + uT Ru)dt, Q ≥ 0, R ≥ 0 (3.4)
0
Where Q and R are the positive-definite Hermitian or real symmetric matrix. The
weighting matrices Q and R finds the scope of the error and the disbursement of the above
mentioned performance index J. From the above equations we get
Z ∞ Z ∞
T T T
J= (x Qx + x K RKx)dt = xT (Qx + K T RK)xdt (3.5)
0 0
1
Where (A, Q 2 ) is detectable and (A−BK) is stable. The aim of the LQR problem
is to produce a control law u = −Kx by minimizing the objective function J with the
solution given by,
K = −R−1 B T P (3.6)

The controller K can be solved using parameter-optimization techniques, we set
d T
xT (Q + K T RK)x = − (x P x) (3.7)
dt
Then we obtain
xT (Q + K T RK)x = −ẋT P x − xT P ẋ = −xT [(A − BK)T + P + P (A − BK)]x (3.8)
Comparing both sides of the above equations,
[(A − BK)T + P + P (A − BK)] = −(Q + K T RK) (3.9)
R = TTT (3.10)
Where T is a non singular matrix, and
AT P + P A + [T K − (T T )−1 B T P ]T × [T K − (T T )−1 B T P ] − P BR−1 B T + Q = 0 (3.11)
The minimization J of with respect to K requires the minimization of
xT [T K − (T T )−1 B T P ]T [T K − (T T )−1 B T P ]x (3.12)
Which this equation is non negative, the minimum occurs when it is zero, or when
T K = (T T )−1 B T P (3.13)
Hence
K = T −1 (T T )B T P = −R−1 B T P (3.14)
Thus the control law as
u(t) = −Kx(t) = −R−1 B T P (3.15)

In which P must satisfy the reduced Riccati equation:
P A + AT P − P BR−1 B T P + Q = 0 (3.16)
The algebraic riccati equation can be solved using the MATLAB command lqr
(linear quadratic regulator).
[K, P, E] = lqr(A, B, Q, R)
Where,
K is the feedback controller optimal gain.
P is the returned solution to the algebraic riccatti equation and
E is a vector containing the eigenvalues (eigenvalues of A − BK).
A. Weight Matrix Selection
The design of LQR problem includes the selection of weighting matrices Q and
R such that the given system has to achieve the desired requirements mentioned earlier.
The values of choosing weighting matrices Q and R will directly effect on the performance
specifications of the system. The best value of this weighting matrices will also be the
solution of the given objective function.
3.2.2 Linear Quadratic Guassian Control
Optimal control, building on the optimal filtering work of Wiener in the 1940s, reached
maturity in the 1960s with what we now call LQG Control [92]. If a controller is designed
using the LQR, and the observer is designed using Kalman filter, the resulting system is
referred to as LQG Control.
The Kalman filter method provides us the procedure for designing observer gains
for Multi Input- Multi Output (MIMO) systems, such that the designed observer gain is

guaranteed to be optimal even in the presence of noise signal. Consider a plant with LTI
state space representation :
ẋ = Ax + Bu + w (3.17)
y = Cx + Du + v (3.18)
The unbiased process noise w and measurement noise v are respectively modeled
as P and Q dimensional unncorrelated white Gaussian process with zero mean and the
covariance matrices are given by:
E{ww0 } = W , E{vv 0 } = V & E{wv 0 } = N .
Where, E{·} is an expectation function operater, W is the exogenous perturbation

vector (simply, defined as process noise matrix) and V is the output disturbance vector
(simply, defined as measurement noise matrix) respectively. The optimal LQG control law
is given by [93]:
u(t) = −kx̂(t) (3.19)
Figure 3.1: Block Diagram of Optimal LQG compensator
The Kalman filter is an optimal estimator when dealing with Gaussian white noise.
Specifically, it minimizes the asymptotic covariance of the estimation error e0 = x − x̂.
limt→∞ E((x − x̂)(x − x̂)0 )

The goal is to regulate the plant output y around zero. The state equation of the
kalman filter can be written as follows:
x̂˙ = Ax̂ + Bu + L(y − C x̂ − Du) (3.20)
Since the Kalman filter is an optimal observer, the problem of Kalman filter is
solved quite similarly to the optimal control problem. For the time invariant problem, the
following algebraic riccati equation results for the optimal covariance matrix [93].
AS + SAT − SC T V −1 CS + BW B T = 0 (3.21)
The algebraic riccati equation can be solved using the specified Kalman filter
MATLAB command lqe (linear quadratic estimator). The Kalman filter optimal gain L,
is given by:
[L, S, E] = lqe(A, B, C, W, V )
Where,
L is the returned kalman filter optimal gain.
S is the returned solution to the algebraic riccatti equation and
E is a vector containing the eigenvalues of the kalman filter (eigenvalues of A−LC).
Finally, combining the separately designed optimal LQR (k) regulator and kalman
filter into an optimal compensator (LQG) as shown in Fig. 3.1.
3.3 Switched Linear Systems
The switching between numbers of subsystems is guided by a switching control law that
takes care of governing the change among these subsystems. In LTI systems, by in-

telligently switching between several of control structures depending on the state value,
improved performance can be achieved compared to individual control structure. Hybrid
or switching systems are characterized by combination of both continuous and discrete
systems and they have created a enormous growth of interest in both systems science &
engineering communities.
A switched-linear system model is as follows:
ẋ(t) = Aσ(t) x(t) (3.22)
The switching signal σ(t) indicates
ẋ(t) = A1 x(t) = if, σ(t) = 1 (3.23)
= A2 x(t) = if, σ(t) = 2
3.3.1 Motivation for switching
The following points mentioned below motivated us to do the research considering switch-
ing system:
• Switching between various structures is an essential feature of many engineering

applications such as power systems.
• Use of intelligent switching techniques in control has argument to give best per-
formance when compared to the performance of a system which does not use any
switching.
• The requirement of switching signal has many reasons. One of them is that, in many
engineering applications, restrictions on the switching signal cannot be specified a
prior.
• Another important motivation for designing switching control strategies is to ensure

robust control performance in the presence of component failure.

• Switching between different optimal controllers for the specified operating condition
can be implemented instead of trying to optimize a single optimal controller for all
the operating conditions.
• In the context of power system, switching between different FACTS controllers can
be designed to combine the properties of both FACTS controllers and to obtain a
new property which is not present in any one of the FACTS controllers.
3.3.2 Switching control Algorithms
The two standard switching control algorithms [73], [74] which was suitably modified and
implemented according to the power system based LTI models in the present research are
briefed here.
A. Switching Algorithm I
The switching boundary vectors F1 & F2 are designed using this switching control
algorithm as follows:
0 0
ẋ(t) = (A + BK1 )x(t) x F1 F2 x > 0
0 0
= (A + BK2 )x(t) x F1 F2 x ≤ 0 (3.24)
1. Firstly, find an alternate controller K2 such that it has n − 1 closed loop real stable
eigenvalues.
2. Secondly, from the closed loop eigenvalues of (A + BK1 ) select a primary controller
K1 such that it has n − 2 common eigenvalues of (A + BK1 ) and the remaining
eigenvalues are complex.
3. To design F1 , multiply the left side eigenvalue polynomials of (A + BK1 ) and select
the coefficients of expanded polynomial in ascending powers of s.

4. To design F2 = [F1 + µω2 ], ω2 is calculated by multiplying the polynomial that is

removed while designing the vector F1 (right side eigenvalue) with other (n − 2) left
eigenvalues by selecting µ < 0.
A (i): Stability of switching algorithm I
The main attention in this subsection is to establish a stabilization via switching

between two feedback controllers. Consider a LTI system
ẋ = Ax + Bu (3.25)
Where, the pair (A, B) is reachable. The task is to search the row vectors K1 , K2 ,
F1 & F2 , in order to achieve exponentially stable for switched system.
0 0
ẋ(t) = (A + BK1 )x(t) x F1 F2 x > 0
0 0
= (A + BK2 )x(t) x F1 F2 x ≤ 0 (3.26)
Firstly, assume that F1 6= γF2 for any γ ∈ R (vectors F1 and F2 which do not
satisfy this constraint implement switching laws which use the matrix A + BK1 only on
the hyper plane F1 x = 0, a measure zero set in Rn ). Under the following assumptions, the
switched system of Eq. 3.26 is globally exponentially stable:
• A+BK1 has n−1 closed loop eigenvalues in the right half plane (at least one of which
is purely real), along with a single, real dominant eigenvalue λ1 with corresponding
right eigenvector v1 .
• F1 = N , where N is the normal vector to the hyper plane containing the n − 1

smallest eigenvalues of A + BK1 .
• F2 is neither a left eigenvector of A + BK1 nor A + BK2 .

• The dominant right eigenvector v1 of A + BK1 satisfies the condition
0 0
v1 N F2 v1 > 0.
• The following two conditions hold:
0
x N 0 F2 (A + BK1 )x ≥ 0 ∀x : F2 x = 0
0
x N 0 F2 (A + BK2 )x ≥ 0 ∀x : F2 x = 0
• One of the following conditions must hold:
1. The matrix A2 has a pair of conjugate symmetric eigenvalues λ˙1 , λ˙2 ∈ C, λ˙1 =λ˙2 ,
0
Im λ˙1 =
6 0 such that the corresponding right eigenvector V˙1 of λ˙1 satisfies the condi-
tion N V˙1 6= 0.
2. The matrix A2 has a pair of real eigenvalues λ˙1 , λ˙2 ∈ R such that the corresponding
right eigenvectors V˙1 and V˙2 satisfy the conditions
0
0
V˙1 N F2 V˙1 < 0
0
0
V˙2 N F2 V˙2 < 0
The above stated assumptions are satisfied then the switching law of the form in
Eq. 3.26 is globally exponentially stable. Proof, corollary & derivations to satisfy the
above assumptions are in [94].
B. Switching Algorithm II
The switching control algorithm II, for finding the switching matrix is as follows:
0
ẋ(t) = (A − BK1 )x(t) x Sx ≤ 0
0
= (A − BK2 )x(t) x Sx > 0 (3.27)

1. Define A1 has asymptotically stable system and A2 to be may or may not be stable.
2. Determine T0 by solving the algebraic Lyapunov Equation: AT1 T0 + T0 A1 = −C T C.
3. Find the switching matrix S from the switching matrix equation: S = −(AT2 T0 +
T0 A2 + C T C).
B (i): Stability of switching algorithm II
In this subsection proof of the asymptotic stability of proposed switching control

is explained:
For the observable system in which sliding not takes place, the proposed switching
control strategy will guarantee the asymptotically stable phenomenon..
Proof: Let us consider the switching takes place ideally with S(x) ≤ 0, the primary
subsystem A1 is in place and when S(x) > 0 the alternate subsystem A2 is used without
any delays.
S(x) = − ≺ x, (AT2 T0 + T0 A2 + C T C)x (3.28)
Let,
AC = γA2 + (1 − γ)A1 (3.29)
Let, γ = 0, 1
When S(x) > 0=Alternate subsystem (γ = 1)=A2 and S(x) ≤ 0=primary sub-
system (γ = 0)=A1 .
Consider, ẋ = AC x with the absence of sliding and the positive definite function
as
V (x) =≺ x, T0 x (3.30)

For Lyapunov stability consider
d ≺ x, T0 x
V̇ =
dt
= ≺ x, (ATC T0 + T0 AC )x
= ≺ x, ATC T0 x + ≺ x, T0 AC x
= ≺ x, (γA2 + (1 − γ)A1 )T0 x + ≺ x, T0 (γA2 + (1 − γ)A1 )x
= ≺ x, γAT2 T0 x + ≺ x, (1 − γ)AT1 T0 x + ≺ x, γT0 A2 x + ≺ x, γT0 A1 x
= −γS(x) − (1 − γ) ≺ x, C T Cx (3.31)
For, an observable system the function −X, C T CX will always be non positive.
When S < 0, γ = 0, the first term γS(X) deletes from Eq. 3.31. When S > 0, γ = 1,
and V̇ is negative. Hence, we can state that the switching system is asymptotically stable.
The proof of sliding along each of the switching surfaces is in [75].
3.4 Optimization of Performance Index Via Switching Control
The aim of the proposed research is to design a optimal switching strategy for power
system with FACTS based controllers such that it would give lesser performance index
J compared conventional controllers (without switching) for the power system prominent
parameters such as rotor angle & rotor speed deviations.
In the present investigations switching between two controllers is considered in an

optimal control, the performance index J of the switching control variable, given in Eq.
3.32, should satisfy for optimal switching.
JSwitch ≤ min(J1 , J2 ) (3.32)
Where, J1 & J2 are the performance index of two subsystems A1 & A2 given in
Eq’s. 3.26 & 3.27, where switching takes place between these two subsystems.

3.4.1 Optimization of JSwitch Using Switching Algorithm I
The switching control algorithm I presented in [74,86], shows that the stabilization of two
LTI systems can be achieved even though one of the subsystem is not stable. But, the
authors instructed that in future optimal switching law can be implemented using the
concepts of these algorithm. By considering the concepts and future direction of these
switching control algorithm an optimal switched feedback control law is proposed for the
LTI systems which finds a controller automatically rather than having to select certain
arbitrary choices as presented in these algorithm. The performance index J defined for
these switching algorithm I is as follows:
Z t
J= (||y||)2 dt (3.33)
0
The primary controller for this switching control algorithm are designed from LQR
based optimal control and the secondary controllers are selected using the pole placement
method by using the concepts of switching control algorithm. Where, the primary con-
troller is asymptotically stable and secondary controller is unstable controller (switching
R∞
algorithm). The performance index of primary controller (LQR) J1 = 0 (xT Qx+uT Ru)dt
will give lesser value depending on the weighting matrices (Q & R). Secondary controller
performance index J2 will be very higher value due to the unstable controller of pole place-
ment method (higher value of Mp & ts ). As approximations J = Mp2 ts can be considered
for this switching algorithm.
Let average value of y(t) when system is in mode 1 be yˆ1 and average dwelling
time in mode 1 be τ1 .
Let average value of y(t) when system is in mode 2 be yˆ2 and average dwelling
time in mode 1 be τ2 .
Due to stabilization algorithm, by design τ2 < τ1 and instantaneous switching en-

sures that yˆ1 τ1 < yˆ2 τ2 < yˆ1 (τ1 +τ2 ). Thus heuristically we argue that JSwitch ≤ min(J1 , J2 ).

Experimental observations conducted in this work are consistent with this result,
although a strong mathematical proof is still required.
3.4.2 Optimization of JSwitch Using Switching Algorithm II
The switching control algorithm II presented in [73,75] demonstrates that optimization is

achieved by minimizing the performance index J (output energy) defined as
Z t
J= (||y||2 )dt (3.34)
o
The proposed switching control strategy should give lesser value of the perfor-
mance index JSwitch when switching measured using Eq. 3.33, compared to the equivalent
performance index of J1 & J2 .
Let, A1 & A2 be the two closed loop subsystems with A1 designed as asymptotically
stable and A2 is designed with not necessarily stable. The result of stable matrix A1 has
been analyzed with the lyapunov equation given in Eq. 3.34.
AT1 P + P A1 = −Q (3.35)
From the above Lyapunov theory, we can also define V = xT P x
By considering the matrix Q = C T C as observable matrix it results in positive

definite solution to the Lyapunov equation in Eq. 3.34. Now, for the present case the
stable matrix A1 can be written as in Eq. 3.35.
T0 A1 + AT1 T0 = −C T C (3.36)
Where T0 is positive definite solution.
The performance index J1 of the primary controller (A1 ) is defined as

Z ∞
J1 = ||y 2 ||dt =≺ x0 , T0 x0 (3.37)
0
Proof: Let for any system
Z ∞
J1 = y T ydt
Z0 ∞
= xT C T Cxdt (y = Cx)
0
Any stable systems starting from an initial condition x0 is given by
x = eA1 t x0
For, the observable system, we can write as
Z ∞
J1 = xT0 eA1 t C T CeA1 t x0 dt
Z0 ∞
= xT0 eA1 t QeA1 t x0 dt
0
Z ∞
= − xT0 eA1 t (T0 A1 + AT0 T0 )eA1 t x0 dt
Z ∞0
d≺ x0 eA1 t , T0 x0 eA1 t
=
0 dt
= ≺ x0 , T0 x0
Hence, for the initial state x0 the cost of using stable system A1 is
J1 =≺ x0 , T0 x0
Now, we will proceed to see the performance index Jswitch when switching takes
place between A1 (Asymptotically stable matrix) & A2 (not necessarily stable matrix).

For an initial state x = ζ, the use of alternative subsystem A2 for some time τ will
be beneficent if
T
α(ξ, τ ) =≺ ξ, eA2 τ (T (τ ) − T0 )eA2 τ ξ > 0 (3.38)
where T (τ ) is the solution, at time τ , of the differential equation and T0 is the

solution of the Lyapunov equation
Ṫ + AT2 T + T A2 = −C T C; t ≥ 0; T (0) = T0
T0 A1 + AT1 T0 = −C T C
Proof: Consider the differential equation in the form,
Ṫ + AT2 T + T A2 = −C T C; t ≥ 0; T (0) = T0 (3.39)
T
Pre multiplying and post multiplying the above Eq. 3.38 by eA2 t and eA2 t respec-
tively,
T T
eA2 t (Ṫ + AT2 T + T A2 )eA2 t = −eA2 t C T CeA2 t
From the above it is clear that
T
d(eA2 t T eA2 t ) T
= −eA2 t C T CeA2 t
dt
For any given τ > 0, quadratic performance of the controlled variable when using
the alternate not necessarily closed loop stable matrix A2 in the interval [0, τ ] and starting
from the initial state ξ0 is

Z τ
T
Jswitch = ≺ ξ0 , ( eA1 t C T CeA1 t )ξ0
0
= ≺ ξ0 , T (0)ξ0 − ≺ x1 , T (τ )x1
where x1 = eA1 τ ξ0 . The performance of the system that switches to primary

subsytem at time τ is given by ≺ x1 , T0 x1 . Assume that the system starts from a initial
state ξ0 and uses the alternate subsystem initially. The cost of the system starting with
alternate subsystem and switching to primary subsystem at some time τ is
Jswitch =≺ ξ0 , T (0)ξ0 − ≺ x1 , T (τ )x1 + ≺ x1 , T0 x1
Hence, the use of alternative subsystem A2 would be helpfull only if there exists a
time τ > 0 such that
≺ x1 , T (τ )x1 − ≺ x1 , T0 x1 =≺ x1 , (T (τ ) − T0 )x1 > 0 (3.40)
For an initial condition ξ ∈ Rn , we can write Eq. 3.39 as
T
α(ξ, τ ) =≺ ξ, eA2 τ (T (τ ) − T0 )eA2 τ ξ > 0 (3.41)
The above Eq. 3.41, implies that the performance of the system can be improved if
the alternate control of subsystem A2 , is applied for τ seconds. From the given above Eq.
3.41, the development of switching function is difficult. Hence, in order to maximize the
function α derivative is implemented such that α at τ > 0 is positive, then we can argue
that there will be a time interval of length τ > 0 where the function α will be positive.
Theorem: For the switched linear system Eq. 3.27, If σ(t) = 2 and if for initial
state ζ, the following condition holds,

dα(ζ, t1 )
|t1 =0 = − < ζ, (AT2 τ0 + τ0 A2 + C T C)ζ > 0
dt1
Where, the switching matrix s is defined as S = −(AT2 τ0 + τ0 A2 + C T C) then for

small values of t1 , the function α will be positive & the alternate control is beneficial.
Proof :
T
α(ζ, t1 ) =< ζ, eA2 t1 (τ (t1 ) − τ0 )eA2 t1 ζ >
T
α(ζ, t1 ) = ζ T eA2 t1 (τ (t1 ) − τ0 )eA2 t1 ζ
T T
α(ζ, t1 ) = ζ T eA2 t1 τ (t1 )eA2 t1 ζ − ζ T eA2 t1 τ0 eA2 t1 ζ
dα(ζ, t1 ) T T
|t1 =0 = ζ T eA2 t1 τ (t1 )eA2 t1 ζ − ζ T eA2 t1 τ0 eA2 t1 ζ
dt1
dα(ζ, t1 )
|t1 =0 = ζ T AT2 τ0 ζ + ζ T A2 τ0 ζ − [ζ T AT2 τ0 ζ + ζ T A2 τ0 ζ]
dt1
dα(ζ, t1 )
|t1 =0 = ζ T [AT2 τ0 + τ0 + A2 ]ζ − ζ T [AT2 τ0 + τ0 + A2 ]ζ
dt1
dα(ζ, t1 )
|t1 =0 = ζ T [−Q]ζ − ζ T AT2 τ0 ζ
dt1
dα(ζ, t1 )
|t1 =0 = − < ζ, (AT2 τ0 + τ0 A2 + C T C) > 0
dt1
Above results are used for designing the switching algorithm II and it also shows,
that the performance of the LTI system is improved when switching JSwitch < min(J1 , J2 )
compared to the conventional LTI systems (without switching).

3.5 Summary
In this chapter, the design of two optimal controllers LQR & LQG (LQR + Kalman
filter) are briefly explained. To know the importance of switched Linear Systems, a brief
motivation is described followed by the two standard switching control algorithms design
is described along with the stability proof. Optimization via switching control is also
discussed in this chapter. By considering the concepts and future direction of proposed
two standard switching control algorithms, an intelligent based novel switching techniques
are proposed for PSS, FACTS and Coordinated Design of PSS & FACTS devices in various
combinations for power system in the next preceding sections.

Chapter 4
MODELING, OPTIMAL CONTROL DE-

SIGN & SWITCHING CONTROL OPER-
ATION OF PSS & UPFC

4 MODELING, OPTIMAL CONTROL DESIGN &

SWITCHING CONTROL OPERATION OF PSS
& UPFC
4.1 Introduction
Chapter 2 provides a general introduction to power system, FACTS devices with compari-
son and linearized Phillips heffron SMIB model is described. Introduction to the switched
linear system along with the optimal control design and two standard switching control al-
gorithms for LTI systems in the current scenario is explained in chapter 3. In summary of
chapter 2 , summarizes the interesting measure of FACTS shows the importance of UPFC.
Chapter 3, suggests that switching between two LTI systems brilliantly has advantageous
because it combines the properties of two LTI system and introduces a new property of
LTI system.
The overall summary of Chapter 2 & 3, motivated to develop switching strategies

for the existing linearized Phillips heffron SMIB model of PSS & UPFC control inputs by
suitably modifying the switching control algorithms proposed in [73] & [74] according to
the requirement of current research.
In the first part of this chapter modeling & analysis of linearized Phillips heffron
model installed with PSS & UPFC is explained followed by the optimal control design.
Switching control operation of PSS & UPFC are experimented with numerical values and
the simulation results are compared.
In the first part of this chapter, switching strategies are developed for uncoor-
dinated design of PSS & UPFC. From, the literature review, it reveals that to improve
the dynamic performance of the power system many researchers have been conducted
the experiments by considering the coordinated design of PSS & UPFC based FACTS
controllers, where they have shown significant improvements in performance compared to

uncoordinated PSS & FACTS based damping controllers. Motivated by the above ob-
servation, switching control techniques are implemented for the simultaneous coordinated
design of PSS and UPFC as second part of this chapter.
4.2 Linearized Model of SMIB with PSS
Power System Stabilizer (PSS) is an additional control which was used for damping oscil-
lations to avoid the power system from becoming unstable. The IEEE excitation system
installed with PSS is as shown in Figure 4.1. The linearized phillips heffron model of
SMIB based PSS input u is represented as shown in Figure 4.2.
Figure 4.1: Block diagram of IEEE excitation system
Figure 4.2: Block diagram of power system installed with PSS

The linearized power system swing and generator internal voltage equations are given by,
δ̇ = ωb (ω − 1) (4.1)
ω̇ = (Pm − Pe − D(ω − 1))/M (4.2)
The internal voltage Ėq0 is given by
0 0 0 0
Ėq = (Ef d − (xd − xd )id − Eq )/Tdo (4.3)
The real power output of the generator is described as
Pe = Vd id + Vq iq (4.4)
The excitation system can be represented by the IEEE type-ST1 system shown in Figure
4.1 and is described by
Ef d = (KA − (Vref − V + UP SS ) − Ef d /TA (4.5)

1
V = (Vd2 + Vq2 ) 2 (4.6)
Vd = xq iq (4.7)
0 0
Vq = Eq − x qid (4.8)
The above linearized equations arranged in the state space form as:

    
˙
∆δ 0 ω0 0 0 ∆δ
    
 ˙  
−K D
−M − Km2
 
 ∆ω   M
1
0   ∆ω 
 =  
 ˙ 0   0
 ∆Eq   − TK0 4 0 − TK0 do 1  
3
T 0 do
  ∆Eq 
   do  
∆E˙ f d − KTAAK5 0 − KTAAK6 − T1A ∆Ef d
 
0
 
 
 0 h i
+

 U
 P SS
 0 
 
KA
TA
The above matrices represents the linearized state space model of SMIB with PSS.
0
Where, the state variables ∆δ=rotor angle deviation, ∆ω=rotor speed deviation, ∆Eq =q-
Axis component deviation and ∆Ef d =field voltage deviation. The control input matrix
UP SS represents the power system stabilizer (PSS) control input.
4.2.1 Experiments using Switching Control Algorithms
The discriminatory of this proposed method, is to introduce switching concepts between

one optimized and other non-optimized feedback controllers by considering the future di-
rections of both the switching control algorithms [73], [74] for the linearized SMIB based
Phillips-Heffron model of PSS. Switching strategy is developed between optimized (pri-
mary) and non-optimized (secondary) feedback controller gains. The primary & secondary
feedback controllers are designed using LQR based optimal control and non-optimal based
pole placement technique respectively to obey the switching rules of both algorithms. The
experimental set up to test the proposed switching model consists of A & B matrices with
the nominal operating conditions [38] & [44].

 
0 377 0 0
 
 
 −.1317 0 −.1104
 
0  T
25
A=
 
 B= 0 0 0 0.06
 −.2356 0 −.463
 
.1667 
 
 
15.47 0 −194.81 −16.667
A. Primary Controller Design:
The Primary controller are designed using the earlier proposed LQR technique for
PSS [44] as follows:
[K, P, E] = lqr(A, B, Q, R, N ) (4.9)
The assumptions in Eq. 4.9 are from the existing approach of [44]. Solving Eq.
4.9, the Riccati equation P is
 
1.8 0 1.5 0
 
 
 0 5329 −34.4 0 
 
P =
 

 1.5 −34.4 1.9 0 
 
 
 
0 0 0 0
Also, the optimal gain control matrix K1 is calculated as

h i
K1 = 0.1217 −0.1090 0.1292 0.0015
The eigenvalues are as follows

 
−14.3968
 
 
−2.8719
 
 
λ=
 

 −0.2429 + 7.0208i
 

 
 
−0.2429 − 7.0208i

B. Secondary Controller Design:
To, satisfy both the switching control algorithms rule, For this research, we (arbi-
trarily) will shift the eigenvalues located at −0.2429 + 7.0208i and −0.2429 − 7.0208i of
the matrix A1 to eigenvalues of +1 & − 1 for the matrix A2 .
Using pole placement technique, place poles at

h i
λ= −14.3968 −2.8719 −1 1
K = place(A, B, P )
h i
K2 = −0.1654 112.7865 −0.7135 0.003
C. Switching Matrices Design:
Using the concepts of switching control algorithms I & II the switching boundary
vectors F1 & F2 and the switching matrix S(∆δ) & S(∆ω) are given by,
h i h i
F1 = 41.3461 58.6148 18.268 1 F2 = 82.6922 34.5377 2 0
 
−0.0020 0.3983 −0.0039 −0.0000
 
 
 0.3983 −3.5214 0.4043
 
0.0037 
S(∆δ) =   ∗ 1.0e + 003
 
 −0.0039 0.4043 −0.0058 −0.0000 
 
 
 
−0.0000 0.0037 −0.0000 0.0000
 
−0.0007 0.1362 −0.0013 −0.0000
 
 
 0.1362 −0.0364 0.1126
 
0.0011 
S(∆ω) = 
 

 −0.0013 0.1126 −0.0017 −0.0000 
 
 
 
−0.0000 0.0011 −0.0000 0.0000

D. Switching Techniques:
The switching techniques for the present research (refer Figure 4.3) as as follows:
ẋ(t) = Aα = (A − BK1 )x(t) (4.10)
= Aβ = (A − BK2 )x(t)
Figure 4.3: Switched Linear system for PSS control input
• Switching Control Algorithm I (Non Lyapunov based):

0 0
ẋ(t) = (A + BK1 )x(t) x F1 F2 x > 0
0 0
= (A + BK2 )x(t) x F1 F2 x ≤ 0
• Switching Control Algorithm II (Lyapunov based):

0
ẋ(t) = (A − BK1 )x(t) x Sx ≤ 0
0
= (A − BK2 )x(t) x Sx > 0
Where the feedback controllers K1 (primary controller) is the optimized and K2

(secondary controller) is the non-optimized feedback controller gains will switch according
to the specified switching conditions in switching control algorithms of I & II.

4.2.2 Result Analysis
The dynamic responses of ∆δ & ∆ω are plotted as shown in Figure 4.4 with the legends
individual optimized LQR controller (K1 ), switching between primary & secondary con-
trollers with switching algorithm I (Sw 2) and switching algorithm II (Sw 1). The ∆δ &
∆ω responses for the pole-placement control method (non-optimized unstable controller)
are shown in Figure 4.5. Figures 4.6 & 4.7 shows the switching signal responses for the
∆δ & ∆ω for both switching algorithm I (Sw 2) as well as switching algorithm II (Sw 1).
Figure 4.4: Rotor angle & Rotor speed deviations responses for individual LQR
& switching control algorithms.
Figure 4.5: Rotor angle & Rotor speed deviation respones for pole placement
control method.

Figure 4.6: Rotor angle & Rotor speed deviation switching signals responses.
Figure 4.7: Rotor angle & Rotor speed deviation switching signal responses.
From Figure 4.4, it is concluded that the switching between optimized (LQR) and
non optimized (pole placement) feedback controllers provides better performance com-
pared to the individual controllers of primary (LQR stable control) and secondary (pole
placement unstable control). Simulation comparison of both switching control algorithms
(refer Table 4.1) reveals that the switching algorithm II, concept provides more better
performance compared to switching algorithm I.

Table 4.1: Comparison of Peak Overshoots (Mp ) & Settling Time (Ts ) for switch-
ing control algorithms
State Switching Switching State Switching Switching

Variables Algorithm I Algorithm II Variables Algorithm I Algorithm II
∆δ -1.3 -1.2 ∆δ 3s 15s
∆ω -0.6 -0.8 ∆ω 5s 16s
4.3 Linearized model of SMIB with UPFC
The idea of using UPFC concept in power system network was first described by L. Gyugyi
in the year 1995. UPFC belongs to the FACTS family, which control the power flow and in
addition to that it also controls the other parameters such as voltage level, line impedance,
phase angle, etc. Hence, it can also be used for power system stabilizing control.
The linearized Phillips heffron model of power system installed with UPFC as
shown in Figure 4.8, developed by H.F. Wang in [44] is given as follows:
˙ = ω0 ∆ω
∆δ
˙ = (−∆Pe − D∆ω)/M
∆ω
∆E˙ 0 = (−∆E 0 + ∆Ef d )/T 0

q q do
1 k A
∆E˙ f d = − ∆Ef d − ∆V
TA TA
Where,
0
∆Pe = k1 ∆δ + k2 ∆Eq + kpe ∆mE + kpδe ∆δE + kpb ∆mB + kpδb ∆δB
0 0
∆Eq = k4 ∆δ + k3 ∆Eq + kqe ∆mE + kqδe ∆δE + kqb ∆mB + kqδb ∆δB
0
∆V = k5 ∆δ + k6 ∆Eq + kve ∆mE + kvδe ∆δE + kvb ∆mB + kvδb ∆δB

Figure 4.8: Block Diagram of Power System installed with UPFC
The above linearized equations arranged in the matrix form as
ẋ(t) = Ax(t) + Bu(t) (4.11)
h i0
0
x(t) = ∆δ ∆ω ∆Eq ∆Ef d
h i0
u(t) = ∆mE ∆δE ∆mB ∆δB
 
 0 ωo 0 0 
 
 
 − k1 D
−M k2
−M
 
 M 0 

A=



 − Tk01 0 − Tk03do 1
 
T 0 do

 do 
 
 
− kTAAk5 0 − kTAAk6 − T1A

 
0 0 0 0
 
 
kpδe k kpδb
 − kMpe −M − Mpb −M
 

B=
 

k k k
 − Tk0qe − Tqδe − T 0qbdo − Tqδb
 
0 0 
 do do do 
 
− kATkAve − kATkAvδe − kATkAvb − kATkAvδb
The above matrices represents the linearized state space model of UPFC. Where,
the state variables are the rotor angle deviation (∆δ), speed deviation (∆ω), q-axis com-
ponent (∆E 0 q ), field voltage deviation (∆Ef d ) and input variables are modulating index
and phase angle of shunt inverter (mE , δE ) and modulating index and phase angle of series
inverter (mB , δB ). A and B represent the state and control input matrices.
4.3.1 Experiments using Switching Control Algorithm
UPFC combines the properties of both shunt and series compensation. The regular prac-
tice for designing the state feedback controller gains for UPFC is from optimal control
theory. But parameters (weighting matrix selections) of optimal controllers are usually
designed based on the trial-and-error approaches and this leads to time consumption
involved in selecting better weighting matrices to design best optimal controller which
damps power system oscillations. To solve this problem, switching strategy method is im-
plemented for UPFC to switch between two optimal Linear Quadratic Regulator (LQR)
controllers to show the improved dynamic performances compared to individual LQR con-
trollers. Switching techniques proposed here are tested with the SMIB Phillips heffron
model installed with UPFC.
A switched-linear system model for the current problem (refer Figure 4.9) is as
follows:

Figure 4.9: General linear switched system for UPFC
The switching signal σ(t) indicates (Aσ(t) )
ẋ(t) = A1 = (A − BK1 )x(t) (4.12)
= A2 = (A − BK2 )x(t)
The two controller gains K1 and K2 matrices represents feedback loop for power
system installed with UPFC. The two feedback controller gains are derived from LQR
optimal control by simply choosing the weighting matrices as identity. Switching algorithm
II, concepts is used for the design of switching signal σ(t).
0
σ(t) = 1 x Sx ≤ 0
0
= 2 x Sx > 0
The switching matrix S is derived from the two closed loop systems master system
(A1 ) & alternate system (A2 ) using switching algorithm II. The experiments has been
carried for all the four UPFC control inputs mE , mB , δE & δB in the following three
stages:

1. Firstly, the two LQR controller gains are derived by selecting the weighting matrices
(without tuning based on any approaches) as Q=I & R=1 (K1 ) and Q=I & R=0.01
(K2 ) for the linearized SMIB based UPFC model.
2. Secondly, switching techniques are implemented to switch between designed two

optimal LQR controllers (master & alternate controllers).
3. Finally, the developed switching model of UPFC are simulated using MATLAB/
SIMULINK
R platform. Results of individual optimal LQR controllers K1 & K2 and
proposed switching between K1 & K2 are compared. To validate the effectiveness of

the proposed technique the results are compared with other optimization techniques
and performance index J is tabulated.
4.3.2 Result Analysis
The proposed experiments of switching control strategy method consists of linearised

Philips Heffron model installed with UPFC described by A and B matrices with the
nominal operating conditions are:
 
0 377 0 0
 
 
 −0.07076 0 −0.0214
 
0 
A=
 

 −0.08322 0 −0.4873 0.1982 
 
 
 
1513 0 −3516 −100
 
0 0 0 0
 
 
 −0.0474 −0.1492 −0.023 −6.612 ∗ 1.0e − 03 
 
B=



 −0.2305 7.53 ∗ 1.0e − 03 −0.056 8.386 ∗ 1.0e − 03 
 
 
4591 −311 1096 −189

The master and alternate feedback controllers of K1 and K2 for all the four con-
trol input matrices B are obtained by using LQR techniques respectively. The proposed
optimal switching strategy S between two vectors of K1 and K2 for the four alternatives
UPFC based damping controllers mE , δE , mB and δB of control matrix B are also given
below.
A. Modulating index of series inverter (mE )
h i
K1 = 1.1660 −48.5192 0.6684 0.9781
h i
K2 = 8.3838 −445.7982 11.3497 9.9743
 
0.0008 −0.0460 0.0011 0.0005
 
 
 −0.0460 2.5498 −0.0597 −0.0289
 

S=
 

 0.0011 −0.0597 0.0014
 
0.0006 
 
 
0.0005 −0.0289 0.0006 −0.0000
B. Phase angle of series inverter (δE )
h i
K1 = −2.931 −101.3626 5.924 −0.6932
h i
K2 = −2.7049 −257.230 −2.184 −9.564
 
0.0000 −0.0079 −0.0004 −0.0004
 
 
 −0.0079
 
1.4343 0.0539 0.0406 
S=
 

 −0.0004
 
0.0539 0.0017 0.0009 
 
 
−0.0004 0.0406 0.0009 −0.0000

C. Modulating index of shunt inverter (mB )
h i
K1 = 2.1271 −149.8195 3.8598 0.9110
h i
K2 = 8.6342 −665.4444 27.2748 9.8972
 
0.0002 −0.0252 0.0008 0.0002
 
 
 −0.0252 2.4708 −0.0820 −0.0215
 

S=
 

 0.0008 −0.0820 0.0023
 
0.0005 
 
 
0.0002 −0.0215 0.0005 −0.0000
D. Phase angle of shunt inverter (δB )
h i
K1 = −1.8283 −289.446 3.1935 −0.5944
h i
K2 = −0.3962 −603.544 6.729 −9.4647
 
0.0005 0.0005 −0.0004 −0.0000
 
 
0.0020 −0.0001 
 
 0.0005 1.6167
S=
 

 −0.0004 0.0020 0.0006 −0.0000 
 
 
 
−0.0000 −0.0001 −0.0000 0.0000
The dynamic response curves for the state variable ∆ω, after the fault at 1 sec-
onds, for all the four control inputs of UPFC mE , δE , mB , and δB are plotted as shown
in the Figures 4.10-4.11 with the legend K1 , K2 and Switch K1 and K2 for the proposed
state feedback optimal switching control. The switching signal responses for all the four
UPFC control inputs are plotted as shown in Figures 4.12 & 4.13. Tables 4.2 & 4.3 shows
the comparison of peak overshoots (MP ) and settling time (TS ) for the two individual

LQR controllers (K1 & K2 ) along with the switching approach (K1 /K2 ). The comparison
R∞
of performance index J = 0 y 2 dt with individual optimal LQR controllers and switching
approach is tabulated in Table 4.4. Table 4.5 shows the comparison of proposed switching
approach with other optimization techniques (without switching) by defining the perfor-
Rt
mance index as J = 0 s MP2 dt.
Figure 4.10: Rotor speed deviation response for mE & δE .
Figure 4.11: Rotor speed deviation response for mB & δB .

Figure 4.12: Switching signal responses mE & δE
Figure 4.13: Switching signal responses for mB & δB
Table 4.2: Comparison of Peak Overshoots (Mp ) for the UPFC control inputs.
UPFC Control K1 K2 Switch K1 / K2

Inputs
mE -6*1.0e-04 -1*1.0e-04 -1*1.0e-04
δE -3*1.0e-03 -3.5*1.0e-04 -3.5*1.0e-04
mB -1.4*1.0e-04 -0.2*1.0e-04 -0.3*1.0e-04
δB -6*1.0e-04 -1*1.0e-04 -1*1.0e-04

Table 4.3: Comparison of Settling Time (Ts ) for the UPFC control inputs.

Inputs
mE 30s 22s 18s
δE 5s 7s 2.5s
mB 15s 18s 15s
δB 12s 19s 10s
R∞
Table 4.4: Comparison of Performance Index (J = 0
y 2 dt) for the UPFC control
inputs.

Inputs
mE 9.386*1.0e-007 9.885*1.0e-009 4.938*1.0e-009
δE 1.554*1.0e-006 4.187*1.0e-008 1.99*1.0e-008
mB 1.317*1.0e-006 2.675*1.0e-008 1.687*1.0e-007
δB 1.505*1.0e-007 1.299*1.0e-008 8.327*1.0e-009

Table 4.5: Comparison of performance index J with Other optimization tech-

niques.
R ts
J= 0
|Mp2 |dt mE δE mB δB
Proposed Optimal 1.7* 1.0e-07 3.06*1.0e-07 1.35*1.0e-08 1.1*1.0e-07
Switching Approach
RCGA [47] 8.67*1.0e-06 6.86*1.0e-06 4.32*1.0e-06 5.04*1.0e-06
PSO [45] 9.8*1.0e-04 1.805*1.0e-03
IPSO [98] 4.32*1.0e-04 1.44*1.0e-03
θ PSO [26] 1.2375*1.0e-03
ANN [99] 2.5*1.0e-06 1.9687*1.0e-06 1.715*1.0e-06 1.4062*1.0e-06
ICA [100] 1.6*1.0e-05 1.2*1.0e-05
Figures 4.10 & 4.11, show the improvements in switching control compared to
non switching control (Individual LQR controllers) for UPFC for all the four control
inputs (mE , δE , mB & δB ). Tables 4.2 & 4.3, show that the switching between two
optimal controllers for the UPFC control input δE provides the robust performance in
peak overshoots and settling time compared to individual optimal controllers of LQR.
In this proposed method, optimization is achieved in two levels. In the first level two
individual controllers are optimized using optimal control theory of LQR. Finally (second
level) an switching control strategy is developed to switch between two individual optimal
controllers, which further optimizes the output energy, which was showed by comparing
the performance index J with individual controllers (K1 & K2 ) along with the switching
approach (K1 /K2 ).
To validate effectiveness of the proposed optimal switching strategy with other

optimization techniques (without switching) the performance index J is defined as J =
R ts
0
|Mp2 |dt and it is is compared with other optimization techniques as shown in Table
4.4. Table 4.5, concludes that the proposed optimal switching strategy provides robust

performance with other non switching optimization techniques.
4.4 Switching Control Design for Simultaneous Coordinated Deign

of PSS And UPFC
In this section, by considering the concepts of state feedback switching law (non lyapunov
based) developed in [74,94], an optimal switched feedback controller is designed which
automatically finds a controller rather than having to make certain arbitrary choices as
presented in [74,94]. This novel approach is presented for the simultaneous coordinated
design of UPFC control input δE (better input from the previous section) and PSS in order
to improve the damping of oscillations in a power system.
4.5 Preliminary Analysis
To ensure the performance improvement of simultaneous coordinated design of PSS &

UPFC a preliminary control analysis is done by considering uncoordinated control of PSS
& UPFC.
In this subsection, a preliminary control analysis is done by designing the feed-

back controllers using optimal control theory of LQR for PSS & UPFC, in order to gain
some knowledge on individual and simultaneous coordinated controlling of PSS and UPFC
control inputs.
Analysis is done in two stages. In the first stage the conventional optimal control
(COC) analysis is done by selecting PSS or UPFC control inputs individually, resulting in
two separate Single Input Single Output (SISO) systems,
ẋ = AX + B̂u
where B̂ = BP SS or B̂ = BU P F C .

The control law is given by
u = −Kx
where, K = KP SS or K = KU P F C are the controller gains for the inputs PSS and
UPFC respectively. Both KP SS and KU P F C were designed by conventional LQR method
with the weighting matrices Q & R to be chosen as identity and state variables rotor angle
deviation (∆δ) and rotor speed deviation (∆ω) were analyzed.
In the second stage COC analysis is done by selecting both PSS and UPFC as the
simultaneous coordinated control inputs resulting in a Multi Input Multi Output (MIMO)
system with,
h i
B= BP SS BU P F C
Now the optimal controller gain K is 2 x 4 matrix for this MIMO model obtained
by LQR algorithm (Q & R to be chosen as identity) for MIMO system. The experimental
set up to test the preliminary analysis, the nominal operating condition values of A, BP SS ,
BU P F C , KP SS & KU P F C along with the coordinated design control K(P SS & U P F C) = Kβ
are given below.
 
0 377 0 0
 
 
 −.07076 0 −.0214
 
0 
A=
 

 −.08322 0 −.4873 .1982 
 
 
 
1513 0 −3516 −100
   
0 0
   
   
−.1492

 0 
  
 
   
BP SS =  BU P F C =
 
 7.533 ∗ 1.0e − 03
  
 0  
   
   
10000 −311
   

h i
KP SS = .8227 −6.1592 .1091 .9901
h i
KU P F C = −2.9311 −101.3626 5.9240 −.6932
 
.1275 −2.0043 −.1212 .9986
Kβ = 
 

−.6452 −56.8949 .2071 −.0307
Analysis results for the ∆δ and ∆ω state variables are presented below in Figure
h i0 h i0
4.14 for the initial conditions x(0) = 1 0 0 0 for ∆δ and x(0) = 0 1 0 0
for ∆ω appropriately. Tables 4.6 & 4.7 shows the comparison of peak overshoots (MP )
and settling time (TS ) for the individual PSS & UPFC control inputs along with the
simultaneous coordinated PSS and UPFC control inputs.
Figure 4.14: ∆δ and ∆ω responses for preliminary analysis
In view of the simulation results of preliminary control analysis, investigation of

Figure 4.14, Table 4.6 and Table 4.7 conludes that the simultaneous coordinated design of
PSS+UPFC control inputs provides better performance compared to the uncoordinated
control of either PSS or UPFC control inputs with respect to peak overshoots and settling
time.

Table 4.6: Comparison of MP for PSS, UPFC and PSS+UPFC control inputs.
State Variables PSS UPFC PSS+UPFC

Rotor Angle
Deviation (∆δ) -1 -0.5 -0.15
Rotor Speed
Deviation (∆ω) -1 -0.6 -0.3
Table 4.7: Comparison of TS for PSS, UPFC and PSS+UPFC control inputs.
State Variables PSS UPFC PSS+UPFC

Rotor Angle
Deviation (∆δ) 30s 2s 1s
Rotor Speed
Deviation (∆ω) 30s 2.1s 1.2s

4.6 Experiments using switching control algorithm
After doing the preliminary control analysis with individual and coordinated ( PSS &
UPFC), a switching based control law is introduced for the simultaneous coordinated PSS
and UPFC for the linearized Phillips-Heffron model. The coordinated design problem of
PSS and UPFC is formulated as switching between two state feedback master & alternate
controller gains and solved by designing the switching control law using the concepts of
switching algorithm I in [74] .
The switched linear model for the current research is as follows:
0 0
ẋ(t) = (A + BK1 )x(t) x F1 F2 x > 0
0 0
= (A + BK2 )x(t) x F1 F2 x ≤ 0
The two controller gains K1 (master controller) and K2 (alternate controller) model
the simultaneous coordinated design of PSS and UPFC gains. In order to satisfy the
switching control algorithm rule (refer section 3.3.2) the two controller gains are designed
using the optimal control theory of LQR and pole placement method respectively,
The proposed switching approach are tested with linearized Phillips Heffron model
of SMIB installed with simultaneous coordinated PSS & UPFC . The primary and sec-
ondary controllers design along with the switching boundary vectors are given below.
A. Master Controller Design
According to switching control algorithm I, the master controller (K1 ), has been
chosen so that the matrix A1 has n−2 common eigenvalues of A2 , and other two eigenvalues
are not real. Optimal control gain K is computed using LQR control Algorithm from Eq.
3.16.
The matrix R and Q in Eq. 3.16 are assumed to be identity. Solving Eq. 3.16.,
the Riccati parameter S is

 
.1688 4.2971 −.0125 0
 
 
 4.2971 381.68 −1.3068 −.0002 
 
P =
 

 −.0125 −1.3068 1.1118
 
0 
 
 
0 −.0002 0 .0001
Also, the optimized controller gain matrix K1 is calculated as
 
0.1275 −2.0043 −.1212 .9896
K1 = 
 

−.6452 −56.8949 .2071 −.0307
The eigenvalues are as follows
 
−10005
 
 
 −4 + 7i
 

λ=



 −4 − 7i
 

 
−1
B. Alternate Controller Design
To obey the switching control algorithm rule the two closed loop matrices A1 and
A2 must satisfy n − 2 common eigenvalues and it should have real n − 1 stable eigenvalues.
To, meet the requirements of this algorithm the eigenvalues located at −4 + 7i and −4 − 7i
of the matrix A1 are moved to the eigenvalues of +1 & − 1 for the matrix A2 .
Using pole placement concept, place poles at
h i
λ= −10005 −1 −1 1
.
K = place(A, B, λ)
 
−4 −2086 −8 −3
K2 = 
 

−135 −67075 −251 −97
C. Switching Matrices Design
Using the concepts of switching algorithm I, the switching boundary vectors F1

F2 are given by,
h i
F1 = 10005 20011 10007 1
h i
F2 = 20010 20012 2 0
To validate the effectiveness of switching control strategy for simultaneous coordinated

design of PSS and UPFC compared to conventional approach (without switching control)
the dynamic responses of rotor angle deviation (∆δ) & rotor speed deviation (∆ω) are
simulated and peak overshoots (MP ), settling time (TS ) and the performnace index (J)
are tabulated.
Figure 4.15: Rotor angle & Rotor speed deviation respones

Figure 4.16: Switching signal response
Table 4.8: Comparison of MP , TS & J for ∆δ.
Controllers MP TS J
K1 -0.15 1.5s 0.1263
Switch K1 / K2 0 0.4s 0.1225
The dynamic plots of space variables ∆δ and ∆ω, are as shown in the Figure 4.15
with the legend K1 (primary LQR controller) and Switch K1 / K2 (proposed switching
control method between primary feedback controller gain k1 & unstable secondary feedback
controller gain K2 ) for the simultaneous coordinated design of PSS+UPFC. Switching
Signal response is also shown in Figure 4.16. The comparison of peak overshoots (Mp ),
R
Settling time (TS ) and the performance index (J = y 2 dt) is tabulated in Tables 4.8 &
4.9.
From the Figure 4.15, Tables 4.8 and 4.9, one can conclude that the proposed
switching approach between one optimized (primary feedback LQR based control) and
other non optimized (secondary feedback pole placement based control) provides better
performance with respect to peak overshoots and settling time compared to optimal LQR
control for the state variables ∆δ and ∆ω. The optimization is also achieved while switch-

Table 4.9: Comparison of MP , TS & J for ∆ω.
Controllers MP TS J
K1 -0.3 1.2s 0.05893
Switch K1 / K2 -0.3 0.5s 0.05754
ing between two feedback controllers compared to individual LQR controller which was
shown by comparing the performance index J (refer Tables 4.8 & 4.9).
4.7 Summary
The effect of switching between uncoordinated and coordinated design for the linearized
Phillips heffron SMIB based PSS & UPFC model are explained in two parts.
In the first part of this chapter, the two feedback controllers are designed one with
respect to LQR (optimized) and other with respect to pole placement (non-optimized)
method for linearized PSS model. Later, the two modified switching control techniques
are proposed for the PSS to switch between optimized & non-optimized feedback controller
gains by satisfying the switching rules of both lyapunov & non-lyapunov based switching
algorithms. The investigation reveals that both the switching control algorithms provides
better performance compared to the individual LQR control. Switching algorithm II
provides more better performance compared to the switching algorithm I with respect to
the prominent parameters of power system such as rotor angle & rotor speed deviations.
The two optimal LQR feedback controllers are designed by tuning the weighting
matrices (Q & R) of LQR for the Phillips heffron SMIB model installed with UPFC.
Switching between two optimal controllers are proposed for all the four control inputs of
UPFC by considering the concepts of lyapunov based switching algorithm II. The sim-
ulation results showed that the switching between two optimal LQR controllers for all
the UPFC control inputs has better performance compared to the individual LQR based

UPFC control inputs. In particular, the UPFC control input δE provides more improve-
ments in damping compared to all other UPFC control inputs. The proposed optimal
switching control strategy also provides better optimization (performance index J is min-
imized) compared to individual LQR controllers as well as other optimization techniques
(without switching control).
In the second part of this chapter, an attempt has been made for introducing
switching concept for the simultaneous coordinated design of PSS and UPFC. Firstly, to
prove the improvement in performance by using coordinated design of PSS and UPFC
compared to uncoordinated design, a preliminary analysis is conducted for the PSS &
UPFC. It reveals that the simultaneous coordinated design of PSS and UPFC provides
improved performance compared to designing PSS & UPFC controller individually in
uncoordinated manner.
Secondly, a switching model is developed for simultaneous coordinated for PSS

and UPFC such that it switches between one optimized and other non optimized feedback
controller gain. In the present approach, the primary feedback controllers are designed
using the optimal control theory of LQR, which finds a automatic feedback controller
as well as optimization is achieved. Based on the pole locations of primary feedback
controller, using the concepts of switching algorithm the secondary feedback controllers
are designed using the pole placement method. The simulation results conclude that the
proposed switching approach for PSS and UPFC provides improved performance compared
to individual LQR with respect to peak overshoots (MP ) and settling time (TS ). The
optimization in output energy is also achieved using switching compared to the technique
of without switching.

Chapter 5
LQR AND LQG BASED OPTIMAL

SWITCHING TECHNIQUES FOR PSS &
UPFC IN POWER SYSTEMS

5 LQR AND LQG BASED OPTIMAL

SWITCHING TECHNIQUES FOR PSS & UPFC
IN POWER SYSTEMS
5.1 Introduction
In the previous chapters 3 & 4, switching techniques are proposed for power system without
disturbances and for individual UPFC control inputs mB , δB , mE & δE coordinated with
& without PSS to achieve better performances in power system. However, in the context of
power system disturbance is one of the factors to destabilize or to reduce the performance
of the system and also in today’s highly complex power system improving the performance
of a system with individual UPFC control input is quite a challenge.
In order to overcome the above challenges, this chapter proposes intelligent novel
switching techniques to switch between two UPFC control inputs with various combina-
tions coordinated with & without PSS to combine the properties of both UPFC control
inputs. Later, the LQG based switching techniques are implemented for UPFC to over-
come the effect of disturbances in the power system.
In the present chapter, optimization is achieved at two levels: Switching rule that
optimizes output energy and individual optimal controllers between which switching takes
place. Hence, in this chapter, four sets of experiments are done. First set of experiments
are without disturbance scenario where switching is done using Linear Quadratic Regula-
tors (LQR’s). Second set is for power systems with disturbances using Linear Quadratic
Gaussian (LQG). The switching control models proposed in this chapter are tested on the
SMIB based linearised Phillips Heffron model of power system installed with UPFC using
MATLAB/SIMULINK
R platform.

5.2 Power system modelling with PSS & UPFC
The state space formulation for the current research can be expressed as follows:
ẋ(t) = Ax(t) + Bu(t), (5.1)
Where, the state variables of the state space model are rotor angle deviation
(∆δ), speed deviation (∆ω), q-axis component deviation (∆E 0 q ) and field voltage devi-
ation (∆Ef d ). A and B represent the system matrix and control input matrices respec-
tively. UPFC has four input variables: modulating index and phase angle of shunt inverter
(mE , δE ) and modulating index and phase angle of series inverter (mB , δB ).
h i0
BP SS = KA
0 0 0 TA
 
0 0 0 0
 
 
 − kMpb −M
kpδb
− kMpe −M
kpδe
 

BU P F C =
 

 − Tk0qb k
− Tqδb − Tk0qedo
k
− Tqδe
 
0 0 
 do do do 
 
− kATkAvb − kATkAvδb − kATkAve − kATkAvδe
The nominal data values used in this chapter for experiments are from [97].
5.3 Switching Control Theory
The LTI switched systems has found many applications in the areas such as aircraft,
robotics, power system, etc. In the context of power system, Switched system has very
broad applications. Rapid development of cities and industries in the developed and
developing countries, increases the demands for energy particularly in electricity.

A switched-linear or hybrid system model is represented as follows:
ẋ(t) = Aσ(t) x(t) (5.2)
The switching signal σ(t) is defined as:
ẋ(t) = A1 x(t) = if, σ(t) = 1 (5.3)
= A2 x(t) = if, σ(t) = 2
Where;
A1 = A − Bα Kα
A2 = A − Bβ Kβ
The design of feedback controllers Kα and Kβ for the switching process are selected
from any of the advanced control theory. For, the current research this two feedback
controllers are designed with the optimization technique of LQR. The two control inputs
Bα & Bβ in the current research are the PSS input and four control inputs of UPFC
(mB , δB , mE & δE ). In addition to selecting this optimized controller, it is proposed to
implement an intelligent switching algorithm that further optimizes performance.
5.3.1 Switching Algorithm
The switching algorithm that stabilizes overall system and also achieves minimization of
output energy is shown in section 3.4.2 [73, 91]. The suitably modified switching algorithm
for the current research are:
1. Let, consider the two closed loop systems A1 & A2 .
2. Determine T0 by solving the algebraic Lyapunov Equation: AT1 T0 +T0 A1 = −C T C

3. Using, A2 define the switching matrix: S = −(AT2 T0 + T0 A2 + C T C)
4. Switching rule for the current research is, σ(t) = 2 if ≺ x, Sx > 0
=1 otherwise
5.4 Experimental Set-up
The proposed novel switching techniques for the various combinations of UPFC control
inputs along with the coordinated PSS control input are experimented by considering the
following cases:
Case I.(i): Switching Between Two Uncoordinated UPFC The control

input matrix B for this case is defined as B = [Bα Bβ ]. The two control inputs of Bα
& Bβ are different UPFC control inputs in various combinations. The switching between
UPFC control inputs are (mB / δB ), (mE / δE ), (mB / mE ) and (δB / δE ). In order for
Kα & Kβ to correspond with their respective Bα & Bβ , they are modified and defined as
shown in Table 5.1.
Table 5.1: UPFC control inputs for switching
h i
Sl. No. B= Bα Bβ Kα Kβ
   
h i
0 KmB 0
1 mB δB kα =   kβ0 =  
0 KδB
   
h i
0 Km E 0
2 mE δE kα =   kβ0 =  
0 KδE
   
h i
0 KmB 0
3 mB mE kα =   kβ0 =  
0 Km E
   
h i
0 K δB 0
4 δB δE kα =   kβ0 =  
0 KδE

Case I.(ii): Switching Between Coordinated PSS & UPFC Previous at-
tempts in the above literature survey have shown that the coordinated design of of PSS
and UPFC control inputs provides better performance compared to uncoordinated control
inputs. The control input matrix B for this case is defined as B = [BP SS Bα Bβ ]. The
three control inputs of BP SS , Bα & Bβ are PSS & different UPFC control inputs in var-
ious combinations. The coordinated control inputs considered for switching in this case
are (P SS + mB / P SS + δB ), (P SS + mE / P SS + δE ), (P SS + mB / P SS + mE ) and
(P SS + δB / P SS + δE ) as shown in Table 5.2. In order for Kα & Kβ corresponding to
their Bα & Bβ they are modified and defined are shown in Table 5.3. The control input
PSS in this case remains constant in all the combinations only UPFC control inputs (mB ,
δB , mE & δE ) are switching according to previously mentioned switching criterion.
Figure 5.1: Block Diagram of the optimal switching LQR compensator
The block diagram of switching between two optimized control inputs using LQR
for both the cases is as shown in Figure 5.1 and it indicates that only one of the optimized
LQR based control input is present at a particular time depending upon the switching
Rt
supervisor to minimize the performance index J = 0 y 2 dt.

Table 5.2: Coordinated PSS & UPFC control inputs for switching
h i
Sl. B = BP SS Bα Bβ
h i
1 P SS mB δB
h i
2 P SS mE δE
h i
3 P SS mB mE
h i
4 P SS δB δE
Table 5.3: Modified controller gains for switching
Sl. No. K K
 α   β 
K K
 P SS   P SS 
0 0
1 Kα =  Km B  Kβ =
   
0 
   
0 KδB
   
K K
 P SS   P SS 
0 0
2 Kα =  KmE  Kβ =
   
0 
   
0 KδE
   
K K
 P SS   P SS 
0 0
3 Kα =  Km B  Kβ =
   
0 
   
0 Km E
   
K K
 P SS   P SS 
0 0
4 Kα =  KδB  Kβ =
   
0 
   
0 KδE

Case II.(i): LQG Based Control without Switching LQR control gives a
good performance when the plant is not corrupted by disturbance. But when disturbance
acts on the system, LQR gives non-zero steady state error. In order to overcome this,
LQG control has been designed which can handle process noise upto 3.5% as well as
measurement noise upto 1%. To show this effectiveness of LQG control over LQR control
under disturbance the simulations are carried by considering the process noise as W =
10 ∗ B 0 ∗ B & measurement noise as V = 0.1 ∗ C ∗ C 0 .
Case II.(ii): LQG Based Control with Switching To show the effectiveness
of the proposed switching techniques the system is subjected to disturbance. In previous
attempts (Jisha Shaji & Aswin et. al 2015) shows that the LQR control has a very
small disturbance rejection capability. In order to overcome this limitation, an LQG
(Combination of LQR + Kalman filter) has been designed which can handle process noise
as well as measurement noise. The switching between two optimized feedback controller
gains using LQG technique is considered in this case as shown in Figure 5.2. Now, the
switching rule σ(t) is given by,
A1 = A − BKα
A2 = A − BKβ
Figure 5.2: Optimal switching LQG based Control
From the Figure 5.2, it reveals that Kalman filter (observer gain) will reject the
state & output noises present in the system and the switching takes place between two

optimized LQR controller gains according to switching supervisor without affecting the
system and shows the improvements in switching compared to the individual LQG con-
trollers.
5.5 Comparison of simulation results for LQR & LQG optimal

switching controllers with disturbances
To show the improvements in switching, simulations are carried for all the above stated
cases in all combinations for state space variable rotor speed deviation of power system
installed with PSS & UPFC (Philips Heffron model). The numerical values of A & B
matrices for the nominal operating conditions and the optimized LQR feedback controller
gains for individual PSS & UPFC (uncoordinated) control inputs are shown below:
 
0 377 0 0
 
 
 −0.0168 0 −0.1696
 
0  h i0
A= BP SS =
 
 0 0 0 1000
 −0.0393 0 −0.484 0.1983 
 
 
 
58.80 0 −333.70 −20
BU P F C =
mB δB mE δE
       
0 0 0 0
       
       
 −0.046   0.17   0.0119   −0.001 
       
       
       
 0.201   0.1501   0.019   −0.0011 
       
       
       
−561.2 0.60 −20.8 −4.80
       

h i
KP SS = 0.3686 −132.6922 12.5499 0.9827
h i
Km B = −0.8668 71.2779 −8.7704 −0.9770
h i
KδB = 2.3957 110.9214 −9.9935 −0.0023
h i
KmE = 0.9062 337.1713 −14.3254 −0.4622
h i
KδE = −0.711 1142.92 −37.911 −0.4112
The optimized coordinated PSS & UPFC LQR feedback controller gains are reprsented
as (Ka = P SS & mB ), (Kb = P SS & δB ), (Kc = P SS & mE ), and (Kd = P SS & δE )
are also shown below:
 
−0.2501 −61.8383 2.3494 0.8588
Ka =  
−0.5956 −92.0354 2.3032 −0.4786
 
0.0560 −1.6193 0.2104 0.9802
Kb =  
0.9079 63.4799 −0.9812 0.0003
 
0.2460 −76.6973 8.4305 0.9816
Kc =  
0.2983 110.6368 −4.0838 −0.0212
 
0.3667 −131.7672 12.4809 0.9827
Kd =  
0.0199 16.7590 −0.7984 −0.0049

The kalman filter observer gain matrix, L, for the UPFC control inputs is as
follows:
0
L(mB ) = 0.0000 300 −31100 1306200
0
L(δB ) = −0.0000 2.9339 −21.8719 256.0367
0
L(mE ) = −0.0000 4.34615 −37.6230 −30.0582
0
L(δE ) = 0.0000 2.6929 −20.6929 246.1281
The switching matrices for all the above cases are also given below.
A. Switch mB /δB :
 
−0.0008 0.2327 −0.0045 −0.0000
 
 
 0.2327 22.5519 −1.2607 0.0003 
 
S=
 

 −0.0045 −1.2607 0.0441 −0.0001 
 
 
 
−0.0000 0.0003 −0.0001 −0.0000
B. Switch mE /δE :
 
−0.0000 −0.0022 0.0000 0.0000
 
 
 −0.0022 −0.9435 0.0267
 
0.0007 
S=
 

0.0267 −0.0005 −0.0000 
 
 0.0000
 
 
0.0000 0.0007 −0.0000 0.0000

C. Switch mB /mE :
 
−0.0001 −0.0064 −0.0004 −0.0000
 
 
5.3531 −0.1581 −0.0024 
 
 0.0064
S=
 

 −0.0004 −0.1581 0.0027 −0.0001 
 
 
 
−0.0000 −0.0024 −0.0001 0.0000
D. Switch δB /δE :
 
0.0001 −0.0053 −0.0005 0.0000
 
 
 −0.0053 −0.8679 0.0184 −0.0001 
 
S=
 

 −0.0005 0.0184
 
0.0029 0.0000 
 
 
0.0000 −0.0001 0.0000 0.0000
E. Switch P SS & mB /P SS & δB :

 
0.0000 0.0146 −0.0000 −0.0000
 
 
1.4303 −0.0061 −0.0022 
 
 0.0146
S=
 

 −0.0000 −0.0061 0.0000
 
0.0000 
 
 
−0.0000 −0.0022 0.0000 0.0000
F. Switch P SS & mE /P SS & δE :

 
0.0000 −0.0014 0.0001 −0.0000
 
 
 −0.0014 −0.6427 0.0129
 
0.0001 
S=
 

0.0000 −0.0000 
 
 0.0001 0.0129
 
 
−0.0000 −0.0001 −0.0000 −0.0000

G. Switch P SS & mB /P SS & mE :
 
0.0000 −0.0031 0.0000 0.0000
 
 
 −0.0031 −0.3631 −0.0096 −0.0022 
 
S=
 

 0.0000 −0.0096 0.0001
 
0.0000 
 
 
0.0000 −0.0022 0.0000 0.0000
H. Switch P SS & δB /P SS & δE :
 
0.0000 −0.0072 0.0000 0.0000
 
 
 −0.0072 −0.9976 0.0077 −0.0000 
 
S=
 

0.0077 −0.0000 −0.0000 
 
 0.0000
 
 
0.0000 −0.0000 −0.0000 0.0000
The dynamic response of the state variable ∆ω for the case I.(i) are plotted as
shown in Figures 5.3 to 5.4. Case I.(ii) are shown in Figures 5.5 to 5.6 and case II.(i) are
shown in Figures 5.7 to 5.8. Figures 5.9 to 5.10, show the dynamic response for case II.(ii)
(switching between two optimized feedback controller gains when the system is subjected
to disturbance). The optimized feedback controllers are designed by tuning the weighting
matrices of LQR as primary controller K1 as (Q = I & R = 1) and secondary controller
K2 as (Q = I & R = 0.01). The switching signal responses for all the cases are also shown
in Figures 5.11-5.14.

Figure 5.3: ∆ω response of case I.(i) (a) & (b)
Figure 5.4: ∆ω response of case I.(i) (c) & (d)
Figure 5.5: ∆ω response of case I.(ii) (a) & (b)

Figure 5.6: ∆ω response of case I.(ii) (c) & (d)
Figure 5.7: ∆ω response of case II.(i) (a) & (b)
Figure 5.8: ∆ω response of case II.(i) (c) & (d)

Figure 5.9: ∆ω response of case II.(ii) (a) & (b)
Figure 5.10: ∆ω response of case II.(ii) (c)
Figure 5.11: Switching signal response of case I.(i) (a) & (b)

Figure 5.12: Switching signal response of case I.(i) (c) & (d)
Figure 5.13: Switching signal response of case I.(ii) (a) & (b)
Figure 5.14: Switching signal response of case I.(ii) (c) & (d)

Table 5.4: Comparison of peak overshoots, settling time & performance index
J for Case I.(i)
Control Peak Settling Performance Index

Rt
Inputs Overshoots Time J = 0 y 2 dt
mB -0.7 10s 0.6175
δB -0.21 1s 0.02599
mE -0.37 4.5s 0.1796
δE -0.6 8s 0.4142
Coordinated mB + δB -0.21 1.1s 0.04689
Coordinated mE + δE -0.38 4.7s 0.2807
Coordinated mB + mE -0.48 5.3s 0.2818
Coordinated δB + δE -0.21 0.9s 0.02622
Switch mB / δB -0.05 1.5s 0.02271
Switch mE / δE -0.37 4.5s 0.2796
Switch mB / mE -0.35 2.5s 0.1648
Switch δB / δE -0.1 1s 0.02557
Simulation was done to control the variable rotor speed deviation (∆ω) using
the four control inputs. The discussions of the experimental results are explained in the
following two scenarios, scenario I (Case I.(i) & Case I.(ii)) are simulation results carried
without disturbance and scenario II (Case II.(i) & Case II.(ii)) are the experimental results
with disturbances.

Table 5.5: Comparison of peak overshoots, settling time & performance index
J for Case I.(ii)
Rt
Control Peak Over- Settling J = 0
y 2 dt
Inputs shoots Time
Coordinated PSS+mB -0.27 2.2s 0.1129
Coordinated PSS+δB -0.21 1.1s 0.04632
Coordinated PSS+mE -0.45 4.8s 0.2884
Coordinated PSS+δE -0.6 7s 0.4982
Switch PSS+mB / PSS+δB -0.07 0.9s 0.02179
Switch PSS+mE / PSS+δE -0.4 4.1s 0.2787
Switch PSS+mB / PSS+mE -0.25 2s 0.1088
Switch PSS+δB / PSS+δE -0.06 1s 0.02497
Table 5.6: Comparison of peak overshoots, settling time & steady state error
for Case II.(i)
Optimal Peak Settling Steady state

Controllers Oversoots Time Error
LQR - mB 1.8 10s 0.6
LQG - mB -1.2 10 0
LQR - δB 0.3 3s 0.1
LQG - δB 0.55 8s 0
LQR - mE 780 6s 0.05
LQG - mE 600 10s 0
LQR - δE 75 10s 22
LQG - δE 60 10s 0

Table 5.7: Comparison of Performance Index J for the Case II.(ii)
Control K1 K2 Switch K1 and K2

Inputs
δB 0.2134 0.1962 0.1638
mE 1.454*1.0e+005 1.372*1.0e+005 1.311*1.0e+005
δE 4664 3475 2877
5.5.1 Scenario I:
Figures 5.3 & 5.4 show the deviation in rotor speed for individual, coordinated & switching
between different UPFC control inputs mB / δB -(a), mE / δE -(b), mB / mE -(c) and δB /
δE -(d), as mentioned in case I.(i). Figures 5.5 & 5.6 show the response of ∆ω for UPFC
control inputs coordinated with PSS input P SS +mB / P SS +δB -(a), P SS +mE / P SS +
δE -(b), P SS+mB / P SS+mE -(c) and P SS+δB / P SS+δE -(d) as mentioned in case I.(ii).
The comparison of peak overshoots (Mp ), settling time (Ts ) in seconds & performance
index J for case I.(i) & case I.(ii) are tabulated in Tables 5.4 & 5.5.
Figures 5.3- 5.4 and Table 5.4 show that the performance is improved when switch-
ing between UPFC control inputs is considered compared to individual as well as coor-
dinated UPFC control inputs with respect to peak overshoots & settling time. The per-
formance index J is also minimized for switching between different UPFC control inputs
compared to without switching. The switching between UPFC control inputs (mB / δB )
provides better optimization (performance index J is minimized) compared to all other
combinations of UPFC control inputs as in Table 5.4 but settling time is more compared
to individual UPFC control input δB . Finally, Switching between (δB / δE ) has improved
performance in peak overshoots, settling time & performance index J.
The simulation results for switching techniques proposed for coordinated PSS &
UPFC control inputs as mentioned in case I.(ii) (shown in Figures 5.5 to 5.6) indicate the

improved performance in all the combinations over case I.(i). The switching between (PSS
& mB / PSS & δB ) provides robust performance in all the three parameters (Mp , Ts & J)
compared to all other combinations (refer Table 5.5).
5.5.2 Scenario II:
The comparison of two optimal controllers (LQR & LQG) responses are shown in Fig-
ures 5.7 to 5.8 as stated in case II.(i). Results related to case II.(ii) (switching between
two feedback LQG controllers) are shown in Figures 5.9 to 5.10. Table 5.6 shows the
comparison of Mp , Ts & steady state error (Ess ) for the LQR & LQG controllers. The
comparison of performance index J for the two optimized (LQR) feedback controllers when
the disturbance is considered in the system as mentioned in case (iv) is shown in Table
5.7.
Results for case II.(i) shown in Figures 5.7 to 5.8 reveals that using the LQG control
under system is subjected to disturbances the steady state error is minimized compared
to conventional LQR control, refer to Table 5.6.
Case II. (ii) Simulation results (Figures 5.9 to 5.10) show the effectiveness of switch-
ing strategy when disturbances are present in the system. The disturbances (process noise
& measurement noise) are applied simultaneously after 1 seconds to the system. Even
though the noise present in the system, the switching between two optimized LQR feed-
back controllers (K1 / K2 ) along with the kalman filter ensure improved performance
compared to the system without switching. Table 5.7, concludes that the performance in-
dex J is minimized in switching between two feedback controllers compared to individual
controllers.

5.6 Summary
This chapter provides new switching techniques for power system subjected to with &
without disturbances by using PSS and UPFC damping control inputs. The experiment is
settled in two cases in case I.(i) & case I.(ii) without disturbances the switching techniques
are proposed for uncoordinated UPFC & Coordinated UPFC with PSS control inputs and
solved by switching between two LQR based different UPFC control inputs. In case II.(i)
& case II.(ii) the process and measurement noises are considered and solved using LQG
technique.
The proposed experimental results of first case reveals that the switching between
coordinated PSS & UPFC control inputs (case I.(ii)) provides better optimization com-
pared to case I.(i). Switching between (PSS & mB / PSS & δB ) provides robust per-
formance compared to all other combinations of switching proposed in case I.(i) & case
I.(ii).
In second case, (with disturbances) using only LQR control introduces non zero
steady state error for the system. In order to overcome this error LQG technique is
proposed as in case II.(i). It concludes that the steady state error is minimized with LQG
control. The effectiveness of switching between two feedback controllers under disturbances
is shown in case II.(ii) and it reveals that switching between two LQG controllers provides
better optimization compared to individual LQG controller (case II.(i)). The switching
between LQG based δB UPFC control input is the robust controller compared to other
UPFC control inputs. This experiment (case II.(ii)) also provides a platform in future
to propose switching techniques as in case I.(i) & I.(ii) when the system is subjected to
disturbances.

Chapter 6
SWITCHING CONTROL TECHNIQUES

VIA OPTIMIZATION FOR UPFC WITH
EFFECT TO OPERATING CONDI-
TIONS IN POWER SYSTEM

6 SWITCHING CONTROL TECHNIQUES VIA

OPTIMIZATION FOR UPFC WITH EFFECT
TO OPERATING CONDITIONS IN POWER
SYSTEM
6.1 Introduction
In previous chapters switching strategies are developed for PSS & UPFC for single operat-
ing conditions. Due to the large variations in load at the receiving end from lower to higher
level, power system network operates at various operating conditions instead of operating
at single condition. In the context of power system, choosing a better UPFC damping
control input is necessary in the scenario of switching between wide range of operating
conditions. In the earlier optimization approaches the regular passion was designing a
single optimal control law for UPFC to operate at all the loading conditions for better
performance, which was near optimal.
In the present chapter instead of that earlier optimal approaches, individual con-
trollers are optimized for each operating conditions and switching techniques are imple-
mented for that individual optimal controllers to select appropriate optimal controllers for
that particular operating conditions in order to achieve overall controller as best perfor-
mance in all the operating conditions. Choosing the better UPFC control input in this
chapter is converted into a problem of switching between operating points hence cover-
ing various operating conditions and solved by implementing the novel switching control
techniques for different operating points using the switching control law [73].

6.2 Problem Statement
The state space formulation of SMIB based UPFC can be defined as in [96]:
ẋ(t) = Ax(t) + Bu(t)
h i0
0
x(t) = ∆δ ∆ω ∆Eq ∆Ef d
h i0
u(t) = ∆mB ∆δB ∆mE ∆δE
 
 0 ωo 0 0 
 
 
 − k1 D
−M k2
−M
 
 M 0  
A=



 − Tk01 0 − Tk03do 1 

T 0 do 

 do
 
 
− kTAAk5 0 − kTAAk6 1
TA
 
0 0 0 0
 
 
 − kMpb −M
kpδb
− kMpe −M
kpδe
 

B=
 

 − Tk0qb k
− Tqδb − Tk0qedo
k
− Tqδe
 
0 0 
 do do do 
 
− kATkAvb − kATkAvδb − kATkAve − kATkAvδe
Constants k1 to k6 represent the system parameters for different operating condi-

tions [96]. The numerical values of data used in this experiment are from [97]:

Table 6.1: Operating Conditions
Conditions Active Power Reative Power

per unit per unit
Normal Load 1 0.015
Light Load 0.3 0.015
Heavy Load 1.1 0.4
1. Light Load
 
0 377 0 0
 
 
 −0.0232 −0.0575
 
0 0 
AL = 
 

 −0.0172 −0.484 0.1983 
 
0
 
 
20.2297 0 −376.98 −20
 
0 0 0 0
 
 
 −0.0088 0.20 0.0065 −0.0019 
 
BL = 
 

0.048 −0.017 −0.0023 
 
 0.24
 
 
−532.7 7.60 43.70 3.60
2. Normal Load
 
0 377 0 0
 
 
 −0.0168 0 −0.1696
 
0 
AN = 
 

 −0.0393 0 −0.484 0.1983 
 
 
 
58.80 0 −333.70 −20

 
0 0 0 0
 
 
 −0.046 0.17 0.0119 0.001 
 
BN = 
 

 0.201 0.1501 0.019 −0.0011 
 
 
 
−561.2 0.60 −20.8 −4.80
3. Heavy Load
 
0 377 0 0
 
 
 −0.0157 −0.157
 
0 0 
AH = 
 

 −0.040 −0.484 0.1983 
 
0
 
 
66.1087 0 −341.103 −20
 
0 0 0 0
 
 
 −0.10 0.20 0.0112 0.0015 
 
BH = 
 

0.02 −0.0009 
 
 0.20 0.10
 
 
−566.6 −10.1 −28.9 4.5
The proposed work was experimented under the following three scenarios.
• Scenario 1: Switching techniques are implemented for light load and normal load.
Where, A1 = AL − BL KL = Light Load.
A2 = AN − BN KN = Normal Load.
• Scenario 2: Switching techniques are implemented for normal load and heavy load.
Where, A2 = AN − BN KN = Normal Load.
A3 = AH − BH KH = Heavy Load.

• Scenario 3: Switching techniques are implemented for light load and heavy load.
Where, A1 = AL − BL KL = Light Load.
A3 = AH − BH KH = Heavy Load.
6.3 Proposed switching control for different operating condi-

tions
In this section, switched linear system modeled as Phillips heffron system, modified switch-
ing techniques for the proposed switching control algorithm are detailed for various oper-
ating conditions.
6.3.1 Switched Linear Control Systems
Brilliantly switching between two LTI subsystems results in improved performance com-
pared to without switching of individual subsystem.
ẋ(t) = Aσ(t) x(t) (6.1)
The switching signal σ(t) indicates
ẋ(t) = Aα x(t) = if, σ(t) = α (6.2)
= Aβ x(t) = if, σ(t) = β
Kα and Kβ are the individual controller gains of two closed loop systems Aα & Aβ
which can be designed using any of the optimization control techniques. In the proposed
switching technique these optimal controller gains are selected from LQR.

6.3.2 Switching Control Techniques
The optimal switching control techniques to cover all the loading conditions as mentioned
in the above scenarios (refer Figure 6.1) are detailed here.
Figure 6.1: Block diagram of proposed switching system
The UL , UN & UH represents the optimal feedback control inputs as shown in

Figure 6.1. Where, kL , kN & kH are the light, normal and heavy loads feedback controller
gains respectively.
In this work, switching control strategies for the above mentioned scenarios are
defined as:
• Scenario 1: α = 1 & β = 2.
• Scenario 2: α = 2 & β = 3.
• Scenario 3: α = 1 & β = 3.
The UPFC control input matrix B (mB , δB , mE & δE ) for various loading condi-
tions can be defined as

h i h i h i
Bu = BL BN Bv = BN BH Bw = BL BH
In order for kL , kN , & kH (feedback controller gains) to correspond to their re-

spective BL , BN & BH (UPFC control inputs), the modified controller gains are defined
0 0 0
as shown in above Table 6.2. Where KL , KN & KH are modified feedback controller gains
of light, normal and heavy loads.
Table 6.2: Modified Controller Gains
Inputs kL to kN to kH to
BL BN BH
   
0 kL 0 0
Bu kL =   kN =  -
0 kN
   
0 kN 0 0
Bv - kN =   kH =  
0 kH
   
0 kL 0 0
Bw kL =   - kH =  
0 kH
The different light, normal and heavy loading conditions feedback controllers gains
(kL / kN ), (kN / kH ) & (kL / kH ) will switch based on the proposed switching control
algorithm II. In three Scenarios 1, 2 & 3 the evaluation matrices AL , AN & AH takes place
as defined in Table 3.
Now, note that Eigenvalues are

0 0
Eig(A1 ) = Eig(AL − Bu kL ) = Eig(AL − Bw kL )
0 0
Eig(A2 ) = Eig(AN − Bu kN ) = Eig(AN − Bv kN )
0 0
Eig(A3 ) = Eig(AH − Bw kH ) = Eig(AH − Bv kH )

Table 6.3: Allowable System Matrix
Scenario (Plant, Controller)

1 (AL , kL ) & (AN , kN )
2 (AN , kN ) & (AH , kH )
1 (AL , kL ) & (AH , kH )
6.4 Simulation results and Comparison of switching techniques

for all the proposed scenarios
The experimental setup to test the proposed switching control method consists of A and
B matrices for various loading conditions (specified in problem statement). To access the
robust UPFC control input under switching conditions the simulations are carried for the
state variables rotor angle deviation (∆δ) & rotor speed deviation (∆ω) for all the four
UPFC control inputs. The numerical values of optimal LQR controller gains for different
loading conditions are shown in Tables 6.4 to 6.7. The switching matrices evaluated from
switching algorithms [73 & 83] are also described here.
Table 6.4: Control Input mB
k k1 k2 k3 k4
kL -0.9631 32.9854 -2.0962 -0.9657
kN -0.8668 71.2779 -8.7704 -0.9770
kH -1.0408 -13.5068 -3.9607 -0.9661

Table 6.5: Control Input δB
k k1 k2 k3 k4
kL 1.0730 63.1563 -4.1411 0.0980
kN 2.3957 110.9214 -9.9935 -0.0023
kH 2.9017 119.4919 -12.0624 -0.0877
Table 6.6: Control Input mE
k k1 k2 k3 k4
kL 0.8250 122.7012 -1.7600 0.6267
kN 0.9062 337.1713 -14.3254 -0.4622
kH 0.7878 353.2582 -15.1213 -0.5512
Table 6.7: Control Input δE
k k1 k2 k3 k4
kL 0.0871 -242.0356 2.7439 0.1122
kN -0.7 1142.9 -37.9 -0.4
kH 24 1033.4 15 0.4

A. Scenario 1
In this case switching between light load and normal load is considered for all the
four UPFC control inputs. Here the switching matrices S1 & S2 are considered for ∆δ &
∆ω respectively.
1. Modulating Index of Shunt Inverter mB :
 
0.0000 0.0054 0.0000 0.0000
 
 0.0054 −1.1316
 
0.0983 0.0079 
S1 =   ∗ 1.0e + 005
0.0983 −0.0014 −0.0001
 
 0.0000 
 
0.0000 0.0079 −0.0001 −0.0000
 
−0.0001 0.0367 −0.0005 −0.0000
 
 0.0367 −7.0012
 
0.5997 0.0486 
S2 =  
 −0.0005 0.5997 −0.0057 −0.0003
 

 
−0.0000 0.0486 −0.0003 −0.0000
2. Phase Angle of Shunt Inverter δB :
 
0.8994 28.3859 −0.2456 −0.0429
 
17.2262 −1.0963
 
 28.3859 929.446 
S1 =  
 −0.2456 17.2262 −17.4735 −0.1815
 

 
−0.0429 −1.0963 −0.1815 −0.0003
 
−0.0000 0.0080 0.0003 −0.0000
 
0.5424 −0.0225 −0.0009 
 
 0.0080
S2 =  
0.0003 −0.0025 −0.0020
 
 0.0000 
 
−0.0000 −0.0009 0.0000 0.0000

3. Modulating Index of Series Inverter mE :
 
−0.0000 0.0020 −0.0000 −0.0000
 
1.9486 −0.0214 −0.0018 
 
 0.0020
S1 = 
  ∗ 1.0e + 005
 −0.0000 −0.0214

0.0001 0.000 
 
−0.0000 −0.0018 0.0000 0.0000
 
−0.0000 −0.0046 0.0001 −0.0000
 
6.6421 −0.1184 −0.0118 
 
 0.0121
S2 =  
 −0.0001 −0.0237
 
0.0004 0.0000 
 
−0.0000 −0.0007 0.0000 0.0000
4. Phase Angle of Series Inverter δE :
 
−0.0001 −0.0116 0.0004 0.0000
 
 −0.0116 0.1178 −0.0010 
 
4.4115
S1 =   ∗ 1.0e + 004
0.1178 −0.0027 −0.0000 
 
 0.0004
 
0.0000 −0.0010 −0.0000 0.0000
 
−0.0001 −0.0094 0.0002 0.0000
 
 −0.0094 0.0802 −0.0006 
 
2.8047
S2 =  
0.0802 −0.0016 −0.0000 
 
 0.0002
 
0.0000 −0.0006 −0.0000 0.0000
B. Scenario 2
In this case switching between normal load and heavy load is considered for all
the UPFC control inputs. Here the switching matrices S3 & S4 are considered for ∆δ &
∆ω respectively.

 
0.0014 0.6764 −0.0221 0.0005
 
 
 0.6764 6.6100 2.7333 0.6113 
S3 =   ∗ 1.0e + 003
 −0.0221 2.7333 −0.2160 −0.0226
 

 
0.0005 0.6113 −0.0226 −0.0001
 
−0.0001 0.0352 −0.0011 −0.0001
 
 
 0.0352 0.6066 0.1337 0.0319 
S4 =  
 −0.0011 0.1337 −0.0073 −0.0008
 

 
−0.0001 0.0319 −0.0008 −0.0000
 
0.1951 3.6658 −1.0800 −0.0293
 
82.8170 −40.1292 −0.7186 
 
 3.6658
S3 =  
 −1.0800 −40.1292 3.5993 −0.0498 
 
 
−0.0293 −0.7186 −0.0498 −0.0011
 
0.0001 0.0058 −0.0002 −0.0000
 
0.2739 −0.0225 −0.0005 
 
 0.0058
S4 =  
 −0.0002 −0.0225
 
0.0004 0.0000 
 
−0.0000 −0.0005 0.0000 0.0000
 
−0.0002 −0.0188 0.0002 −0.0000
 
 −0.0188 −1.2916 −0.0028 −0.0009 
 
S3 =   ∗ 1.0e + 003
 0.0002 −0.0028
 
0.0010 0.0000 
 
−0.0000 −0.0009 0.0000 0.0000

 
−0.0000 −0.0010 0.0000 0.0000
 
 −0.0010 −0.1370 0.0009 −0.0000 
 
S4 = 



 0.0000 0.0009 0.0000 0.0000 
 
0.0000 −0.0000 0.0000 0.0000
 
0.0002 0.0110 −0.0000 0.0000
 
 0.0110 −1.3461
 
0.0578 0.0008 
S3 =   ∗ 1.0e + 004
 −0.0000 0.0578 −0.0018 −0.0000
 

 
0.0000 0.0008 −0.0000 −0.0000
 
0.0001 0.0126 −0.0001 0.0000
 
 0.0126 −0.7049
 
0.0406 0.0006 
S4 =  
 −0.0001 0.0406 −0.0011 −0.0000
 

 
0.0000 0.0006 −0.0000 −0.0000
C. Scenario 3
In this case switching between light load and heavy load is considered. Here the
switching matrices S5 & S6 are considered for ∆δ & ∆ω respectively.
 
0.0001 0.1969 −0.0013 0.0000
 
 
 0.1969 5.1773 0.999 0.1942 
S5 =   ∗ 1.0e + 004
 −0.0013 0.9999 −0.0163 −0.0016
 

 
0.0000 0.1942 −0.0016 −0.0000

 
−0.0002 0.1190 −0.0011 −0.0001
 
 
 0.1190 3.2348 0.6163 0.1189 
S6 =  
 −0.0011 0.6163 −0.0069 −0.0007 
 
 
−0.0001 0.1189 −0.0007 −0.0000
 
0.0016 0.0482 −0.0012 −0.0001
 
1.4648 −0.0173 −0.0024 
 
 0.0482
S5 =   ∗ 1.0e + 003
 −0.0012 −0.0173 −0.0226 −0.0003 
 
 
−0.0001 −0.0024 −0.0003 −0.0000
 
0.0000 0.0168 0.0003 −0.0000
 
1.0079 −0.0549 −0.0018 
 
 0.0168
S6 =  
0.0003 −0.0549 −0.0024
 
 0.0000 
 
−0.0000 −0.0018 0.0000 0.0000
 
−0.0000 0.0014 −0.0000 −0.0000
 
2.0645 −0.0274 −0.0021 
 
 0.0014
S5 =   ∗ 1.0e + 005
 −0.0000 −0.0274
 
0.0002 0.0000 
 
−0.0000 −0.0021 0.0000 0.0000
 
−0.0001 0.0034 0.0001 −0.0000
 
0.0034 14.0540 −0.1828 −0.0145 
 

S6 =  
0.0001 −0.1828
 
 0.0012 0.0001 
 
−0.0000 −0.0145 0.0001 0.0000

 
0.0003 0.0579 −0.0001 0.0000
 
 
 0.0579 2.8394 0.25260.0013 
S5 =  
 −0.0001 0.2526 −0.0052 −0.0000 
 
 
0.0000 0.0013 −0.0000 −0.0000
 
0.0001 0.0379 −0.0002 −0.0000
 
 
 0.0379 1.9868 0.1671 0.0009 
S6 =  
 −0.0002 0.1671 −0.0032 −0.0000 
 
 
−0.0000 0.0009 −0.0000 −0.0000
The dynamic response curves for the state variables ∆δ & ∆ω are plotted for all the
three scenarios as stated above are shown in Figures 6.5 to 6.7. To show the improvements
in switching for scenarios the individual responses of light (A1 ), normal (A2 ) & heavy (A3 )
loads are also plotted as shown in Figures 6.2 to 6.4.The switching signal responses for all
the three scenarios of four UPFC control inputs are plotted in Figures 6.8 to 6.19.
Figure 6.2: ∆δ & ∆ω responses for Light Load

Figure 6.3: ∆δ & ∆ω for Normal Load
Figure 6.4: ∆δ & ∆ω responses for Heavy Load
Figure 6.5: ∆δ & ∆ω responses for scenario 1

Figure 6.8: Switching signal responses for ∆δ for scenario 1

Figure 6.10: Switching signal responses for ∆ω for scenario 1




Rt
The performance index J = o
||y||2 dt are also tabulated for all the three scenarios
and for individual loads with respect to four UPFC control inputs.
Table 6.8: Peak Overshoots comparison for ∆δ
Control A1 A2 A3 Switch Switch Switch

inputs Light Normal Heavy A1 /A2 A2 /A3 A1 /A3
Load Load Load Scenario 1 Scenario 2 Scenario 3
mB -0.8 -0.5 -0.78 -0.5 -0.6 -0.7
δB -0.05 0 0 -0.05 0 0
mE -0.2 -0.15 -0.15 -0.15 -0.15 -0.15
δE -0.8 -0.45 -0.7 -0.4 MDO MDO
Table 6.9: Settling Time in seconds comparison for ∆δ

mB Above 10 10 Above 10 9 8.1 9.2
δB 1.8 3.3 3.5 1.5 3 1.6
mE 4.2 4.5 4.5 4 3.8 4
δE Above 10 7.8 8.2 7 MDO MDO
The above Figures 6.2 - 6.7, shows the responses of state space variable deviations
for ∆δ & ∆ω for the three individual loads as well as above stated three scenarios. From the
simulation results of three scenarios, it reveals that the UPFC control input δB provides
better performance compared to all other UPFC control inputs of mB , mE & δE with
respect to peak overshoot and settling time as shown in Tables 6.8 -6.11. In Scenarios 2

Table 6.10: Peak Overshoots comparison for ∆ω

mB -0.75 -0.7 -0.75 -0.6 -0.65 -0.7
δB -0.25 -0.22 -0.15 -0.15 -0.13 -0.1
mE -0.4 -0.35 -0.35 -0.3 -0.3 -0.3
δE -0.85 -0.62 -0.57 -0.58 MDO MDO
Table 6.11: Settling Time in seconds comparison for ∆ω

mB Above 10 9 Above 10 8.6 9 Above 10
δB 1.2 1.1 1.2 0.6 0.4 0.5
mE 3.5 4.5 4.5 3.5 3.8 3.5
δE Above 10 8 7.5 7.2 MDO MDO

Table 6.12: Performance Index J comparison for ∆δ

mB 1.258 0.478 0.7692 0.478 0.5001 0.6762
δB 0.1076 0.09847 0.09674 0.0854 0.09522 0.07869
mE 0.2062 0.3813 0.4206 0.3422 0.3793 0.3732
δE 1.327 0.4387 0.5156 0.4 MDO MDO
Table 6.13: Performance Index J comparison for ∆ω

mB 1.339 0.6175 0.7337 0.5302 0.553 0.719
δB 0.04041 0.02599 0.01937 0.02544 0.01937 0.01936
mE 0.1413 0.1796 0.2048 0.1774 0.1734 0.192
δE 1.334 0.4142 0.334 0.386 MDO MDO

& 3, for the control input δE induces more damped oscillations (MDO), hence it is not
shown in the Figures 6.7 - 6.8.
Switching techniques via optimization is implemented in this chapter for all the
loading conditions based on scenarios. Figures 6.5-6.7 & Tables 6.8-6.11, shows the per-
formance improvements in switching compared to three individual loads. When system
switches based on the scenarios of different loads some optimization in the system is also
achieved, which was showed by comparing the performance index J with individual three
loads A1 , A2 & A3 as in Tables 6.12 & 6.13. From the tables, 6.12 & 6.13, it concludes
Rt
that the performance index J = o ||y||2 dt (output energy) is minimized in almost all the
four control inputs mB , δB , mE & δE .
6.5 Summary
In this chapter, optimal switching control has been proposed for UPFC FACTS device
to cover all the operating conditions in power system. Today’s power system network,
load is not constant as it will vary enormously from low to high. In such a situation,
using a single optimal control law to operate for all the loading conditions is always not
optimal. In this chapter for this problem, a switching concept is introduced for certain
operating conditions based on three loads (light, normal & heavy). The switching control
law presented in this chapter are implemented for all the UPFC control inputs (mB , δB ,
mE & δE ) based on three scenarios and experimental results demonstrates that the control
input δB provides improved performance with respect to peak overshoot and settling time
compared to all other UPFC control inputs. Switching between three operating conditions
as in above defined scenarios also shows the optimization achieved via switching control,
which was showed by comparing the performance index J (output energy is minimized)
with respect to individual three loads.
This chapter provides a basic knowledge in future to design a switched optimized

robust UPFC based damping controller for multi machine networks to damp oscillations in

power system, when the load is continuously varying between light/normal/heavy, instead
of designing single optimized UPFC damping control, which was not optimal for all the
operating (loading) conditions.

Chapter 7
SWITCHING CONTROL OF MULTI-

LQRs FOR UPFC AS APPLIED TO
LOADING CONDITIONS IN POWER
SYSTEM

7 SWITCHING CONTROL OF MULTI- LQRs

FOR UPFC AS APPLIED TO LOADING
CONDITIONS IN POWER SYSTEM
7.1 Introduction
The LQR feedback controllers designed in chapter 6, for switching scenarios of different
loading conditions are with single LQR (Bryson Approach). In this chapter multi LQR’s
(Bryson, Boudarel & Multistage) approaches are designed for light, normal & heavy loads
of three operating conditions as preliminary analysis. Later, switching control strategy
is developed to switch between two best LQR’s from preliminary analysis to select ap-
propriate controller into feedback at different operating conditions (switching between
Multi-LQR feedback controller gains).
7.2 Preliminary Analysis
Initial Analysis of the optimal control has been done for the phase angle shunt inverter δB
robust UPFC control input [1-Dissemination]. The data values of all the loading conditions
used in this experiment are from [97]:
1. Light Load
   
0 377 0 0 0
   
   
 −0.0232 0 −0.0575
   
0   0.20 
AL =  BL = 
   
 
 −0.0172 0 −0.484 0.1983 
   
 0.048 
   
   
20.2297 0 −376.98 −20 7.60

2. Normal Load
   
0 377 0 0 0
   
   
 −0.0168 0 −0.1696
   
0   0.17 
AN =  BN = 
   
 
 −0.0393 0 −0.484 0.1983 
   
 0.1501 
   
   
58.80 0 −333.70 −20 0.60
3. Heavy Load
   
0 377 0 0 0
   
   
 −0.0157 −0.157
   
0 0   0.20 
AH =  BH = 
   
 
 −0.040 −0.484 0.1983 
   
0  0.10 
   
   
66.1087 0 −341.103 −20 −10.1
For, these three different loading conditions in power system the closed loop control
system (Aα , Aβ & Aγ ) are given by
Aα = AL − BL KL (7.1)
Aβ = AN − BN KN (7.2)
Aγ = AH − BH KH (7.3)
Where, the feedback controller gains (KL , KN & KH ) are derived form opti-
mal control theory of LQR by tuning the weighting matrices with different approaches
(Bryson,Boudarel & Multistage). The analysis is done by designing three different op-
timal controllers of LQR. For the sake of clarity, three different approaches of LQR are
explained below:

7.2.1 Bryson Rule
Bryson rule is developed in 1975, where the weighting matrices Q & R are to be chosen as
 
1 0 0 0
 
 
 
 0 1 0 0  h i
Q= R=
 
 1
 
 0 0 1 0 
 
 
0 0 0 1
By applying this rule for the design of LQR the optimized feedback controllers
obtained for different loading conditions are:
• Light Load:
h i
KL = 1.0730 63.1563 −4.1411 0.0980
• Normal Load:
h i
KN = 2.3957 110.9214 −9.9935 −0.0023
• Heavy Load:
h i
KH = 2.9017 119.4919 −12.0624 −0.0877
7.2.2 Bouderal Rule
Boudarel rule is developed in 1964, where the weighting matrices Q & R are to be chosen
as
Q = C0 ∗ C R = B0 ∗ B

From this rule LQR is designed the optimized feedback controllers are:
• Light Load:
h i
KL = 0.0612 15.1679 −0.1256 −0.0010
• Normal Load:
h i
KN = 1.4728 80.8967 −0.9352 −0.0051
• Heavy Load:
h i
KH = 0.0835 17.7889 −0.3727 −0.0029
7.2.3 Multistage Rule
This technique of designing LQR is given by R K Pandey in 2010. The design procedure
is as follows:
1. 1st stage: In this stage the LQR is designed using Bryson based LQR.
[k1, s, e] = lqr(A, B, Q, R)
2. 2nd stage: Choose Q1 & R1 matrices as

 
10 0 0 0
 
 
 
 0 1 0 0  h i
Q1 =  R1 =
 
 1
 
 0 0 1 0 
 
 
0 0 0 1
Select, A1 = A − (B ∗ k1)

[k2, s, e] = lqr(A1, B, Q1, R1)
3. 3rd stage: Choose Q2 & R2 matrices as

 
100 0 0 0
 
 
 
 0 1 0 0  h i
Q2 =  R2 =
 
 1
 
 0 0 1 0 
 
 
0 0 0 1
[k3, s, e] = lqr(A2, B, Q2, R2)
4. 4th stage: Choose Q3 & R3 matrices as

 
1000 0 0 0
 
 
 
 0 1 0 0  h i
Q3 =  R3 =
 
 1
 
 0 0 1 0 
 
 
0 0 0 1
[k4, s, e] = lqr(A3, B, Q3, R3)
After, the simulations carried for all the stages (Figures 7.1-7.3), it was concluded
that 4th stage multistage LQR provides better performance compared to the remaining
stages of multistage LQR. The optimized feedback controllers of 4th stage for different
operating conditions are:

• Light Load:
h i
KL = 31.4959 345.4671 −0.1234 0.0230
• Normal Load:
h i
KN = 24.4860 208.4555 0.5622 0.0216
• Heavy Load:
h i
KH = 24.5182 195.4364 0.3714 0.0126
7.2.4 Simulation Results of Preliminary Analysis
To select the better optimized feedback controllers for the specified operating conditions,
simulations are carried for the three optimal LQR’s. Figures 7.4-7.6, shows the responses
of deviations in rotor angle (∆δ) & rotor speed (∆ω) for light normal and heavy loads.
Tables 7.1-7.3, shows the comparison of peak overshoots (MP ) and settling time (TS ).
Figure 7.1: ∆δ & ∆ω responses for light load

Figure 7.2: ∆δ & ∆ω responses for normal load
Figure 7.3: ∆δ & ∆ω responses for heavy load
7.2.5 Discussion on preliminary Analysis
Figures 7.4-7.6, shows the digital simulation results of rotor angle deviation (∆δ) & rotor
speed deviation (∆ω) for light, normal & heavy loading conditions with three different
approaches of LQR. Tables 7.1-7.3, shows the comparison of Mp & Ts for all the loading
conditions with three optimal controllers.
Figures 7.4-7.6 and Tables 7.1-7.3, reveals that for the light and heavy load operat-
ing conditions Bryson LQR provides good response for peak overshoots (Mp ) & Multistage
LQR provides robust performance for settling time (Ts ). In normal load operating condi-
tions also the Bryson LQR is better for peak overshoots but for the settling time Boudarel

Table 7.1: Comparison of Peak Overshoots (Mp ) & Settling time (Ts ) in seconds
for ∆δ and ∆ω with different LQR approaches for light load.
Optimal LQR control Mp (∆δ) Ts (∆ω) Mp (∆δ) Ts (∆ω)

Bryson rule -0.001 1s -0.25 0.9s
Bouderal rule -0.26 3.5s -0.38 3.5s
Multistage rule -0.25 0.4s -0.35 0.45s
for ∆δ and ∆ω with different LQR approaches for normal load.

Bryson rule 0 1.5s -0.2 0.6s
Bouderal rule -0.005 0.75s -0.2 0.55s
Multistage rule -0.4 1s -0.5 0.7s
for ∆δ and ∆ω with different LQR approaches for heavy load.

Bryson rule 0 2s -0.1 0.5s
Bouderal rule -0.35 4s -0.4 4.5s
Multistage rule -0.4 0.4s -0.45 0.5s

Figure 7.4: ∆δ & ∆ω responses for light load
Figure 7.5: ∆δ & ∆ω responses for normal load
LQR provides robust control.
The overall results of preliminary analysis reveals that one LQR is better for peak
overshoot while other LQR is better for settling time. This situation motivated to design an
switching control strategy to achieve optimization between peak overshoots and settling
time. From, the summary of preliminary analysis the two better optimized feedback
controllers selected for the implementation of Multi LQR switching control (master &
alternate controllers) for different operating conditions are as shown in Table 7.4.

Figure 7.6: ∆δ & ∆ω responses for heavy load
Table 7.4: Optimized feedback controllers for different loads
Operating condition Master Control Alternate Control

Light Load Bryson LQR Multistage LQR
Normal Load Bryson LQR Bouderal LQR
Heavy Load Bryson LQR Multistage LQR
7.3 Multi-LQRs Switching Control
The first stage of switching is developed in this section between multi-LQR’s (selected
from the preliminary analysis) as part I.
7.3.1 Switched Linear System
The model for the current switched linear system is as follows (refer Figure 7.7):
ẋ = Aσ x(t) (7.4)
The switching element σ(t) defined as

ẋ(t) = A1 x(t) = if, σ = 1 (7.5)
= A2 x(t) = if, σ = 2
Where;
A1 = A − BK1 (7.6)
A2 = A − BK2 (7.7)
Figure 7.7: General switched linear system
For, the closed loop subsystems A1 & A2 the optimal controller gains K1 and K2 are
derived from any optimization techniques. In the current research these optimal feedback
controller gains are selected from the preliminary analysis of previous section (optimal
control theory of LQR). In addition to selecting these optimized feedback controllers, it
is proposed to implement an intelligent switching control algorithm proposed in [73 & 83]
and section 3.4.2 that further improves the performance in peak overshoots and settling
R∞
time along with further optimizes by minimizing the performance index J = 0 y 2 dt.
7.3.2 Switching Control Strategies
The switching control strategies are developed for all the three operating conditions (light,
normal & heavy loads) are experimented by considering the following cases:

A. Case I:
Light Load: From, the preliminary optimal control analysis Bryson & Multistage LQR
optimal feedback controllers provides better performance in peak overshoots & settling
time appropriately. Switching between these two optimal feedback controllers are proposed
for this case. The closed loop systems from Eq. 7.6 & 7.7 of K1 & K2 are the feedback
controller gains of Bryson and Multistage LQR.
B. Case II:
Normal Load: The two closed loop feedback controller gains for this case are Bryson
(K1 ) and Boudarel (K2 ) LQR (refer preliminary analysis).
C. Case III:
Heavy Load: In this case the preliminary analysis of three optimal controllers suggest to
switch between Bryson (master control K1 ) and Boudarel (alternate control K2 ) optimal
based LQR controllers for the proposed switching control strategy.
7.3.3 Simulation Results of Multi-LQR Switching
To, validate the effectiveness of switching between two optimal controllers compared to
individual optimal controllers simulations are carried for the state variables rotor angle
deviation (∆δ) and rotor speed deviation (∆ω) for the operating conditions (light, normal
& heavy loads). The switching matrices for deviations in rotor angle (S∆δ ) & rotor speed
(S∆ω ) for the different cases are:
Case I: Light Load
 
17.9 302.6 37.1 −0.0000
 
 
−1.8
 
 302.6 2284.9 207.7 
S(∆δ) =
 

−0.2
 
 37.1 207.7 14.7 
 
 
−0.0000 −1.8 −0.2 −0.0000

 
18.9 302.6 37.1 −0.0000
 
 
−1.8
 
 302.6 2283.9 207.7 
S(∆ω) =
 

−0.2
 
 37.1 207.7 14.7 
 
 
−0.0000 −1.8 −0.2 −0.0000
Case II: Normal Load
 
−0.4738 −11.5982 1.1880 −0.0103
 
 
 −11.5982 −253.128 1.1789 −0.3246
 

S(∆δ) =
 

 
 1.1880 1.1789 22.3284 0.0909 
 
 
−0.0103 −0.3246 0.0909 −0.0001
 
0.0000 −0.0037 −0.0002 0.0000
 
 
 −0.0037 −0.2679 0.0374
 
0.0000 
S(∆ω) =
 

 −0.0002 0.0374 0.0018 −0.0000 
 
 
 
0.0000 0.0000 −0.0000 0.0000
Case III: Heavy Load
 
11.0451 87.9403 32.2711 0.3364
 
 
 87.9403 −17.5843 85.6325 0.6144
 

S(∆δ) =
 

 
 32.2711 85.6325 53.4024 0.4902 
 
 
0.3364 0.6144 0.4902 0.0043

 
−0.0008 0.0884 0.0021 0.0000
 
 
 0.0884 −0.0176 0.0879 0.0006
 

S(∆ω) =
 

 
 0.0021 0.0879 0.0049 0.0000 
 
 
0.0000 0.0006 0.0000 0.0000
Figure 7.8, shows the dynamic response of state variables (∆δ) & (∆ω) for the case
I. Case II, responses are shown in Figure 7.9 followed by case III dynamic plots in Figure
7.10. Switching signal responses for all the three cases are shown in Figures 7.11-7.13.
Tables 7.5- 7.10, displays the comparison of Mp & Ts for all the cases. To, highlight the
R∞
switching concept performance index J = 0 y 2 dt are also tabulated in tables 7.11-7.13.
Figure 7.8: ∆δ & ∆ω responses for case I
for ∆δ for case I
Optimal LQR control Mp Ts

Bryson rule -0.001 1s
Multistagel rule -0.25 0.4s
Switch Bryson / Multistage rule -0.05 0.3s

for ∆ω for case I

Bryson rule -0.25 0.9s
for ∆δ for case II

Bryson rule 0 1.5s
Bouderal rule -0.005 0.75s
Switch Bryson / Bouderal rule -0.005 0.7s
for ∆ω for case II

Bouderal rule -0.2 0.55s
Switch Bryson / Bouderal rule -0.15 0.4s

for ∆δ for case III

Bryson rule 0 2s
for ∆ω for case III

R∞
Table 7.11: Comparison of performance index J = 0
y 2 dt for case I
Optimal LQR control ∆δ ∆ω

Bryson rule 0.1076 0.04041
Multistagel rule 0.03648 0.0159
Switch Bryson / Multistage rule 0.02494 0.0115

Figure 7.9: ∆δ & ∆ω responses for case II
Figure 7.10: ∆δ & ∆ω responses for case III
7.3.4 Discussion on multi-LQR switching control strategy
Simulation was done to control the state variables rotor angle deviation (∆δ) and rotor
speed deviation (∆ω) using the UPFC control input δB with switching control concept
for all the operating (light, normal & heavy loads) conditions. The discussions of the
experimental set up of switching control strategy are explained for three stated cases
(light, normal & heavy).
A. Case I
Figure 7.8 and Tables 7.9 & 7.10, shows the response and comparison of Mp & Ts for case I
and it reveals that the switching between Bryson and Multistage LQR optimal controllers
provides better performance with respect to peak overshoots and settling time compared

Figure 7.11: switching signal responses for case I
Figure 7.12: switching signal responses for case II
to individual Bryson and Multistage optimal LQR.
B. Case II
Case II, dynamic response curves and comparisons of Mp & Ts are shown in Figure 7.9
and Tables 7.11-7.12. It concludes that switching between two optimal LQR controllers
(Bryson & Boudarel) provides robust output performance compared to without switching
control (individual optimal controllers).
C. Case III
The dynamic plots and comparison table (Mp & Ts ) of case III, are shown in Figure 7.10
and Tables 7.13-7.14. It revels that switching control strategy between two LQR (Bryson
& Multistage) feedback controllers shows improvements in peak overshoots and settling

Figure 7.13: switching signal responses for case III
R∞
y 2 dt for case II

Bryson rule 0.09846 0.02598
Bouderal rule 0.105 0.03633
Switch Bryson / Bouderal rule 0.08613 0.02357
time compared to without switching control strategy of LQR controllers.
In, all the cases the output energy is minimized in switching between two optimal
R∞
feedback controllers which was showed by estimating the performance index J = 0 y 2 dt
(refer Tables 7.15-7.17) with individual optimal feedback controllers (without switching).

R∞
y 2 dt for case III

Bryson rule 0.09674 0.01938
Multistagel rule 0.02754 0.02139
Switch Bryson / Multistage rule 0.02389 0.006771
7.4 Summary
In this chapter, UPFC FACTS device has been designed for the power system with multi
operating conditions in two parts.
Simulation results of preliminary analysis on three LQR (Bryson, Boudarel & Mul-
tistage) based optimal controllers has been carried out to select the two better optimized
feedback controllers for the specified operating conditions. According to the simulation
results, analysis and comparison the optimized feedback controllers selected for light load
are Bryson and multistage LQR. Bryson & Boudarel LQR are selected for normal load
operating condition. For, heavy loads the Bryson & Multistage LQR are selected. These
optimal feedback controllers are selected in such a way that one is better for peak over-
shoots and other is better for settling time. After selecting the better optimized feedback
controllers from the preliminary optimal control analysis, switching control strategy is
implemented to switch between selected optimized feedback controllers for the specified
operating conditions to achieve optimization between peak overshoots and settling time
in order to obtain overall improved output performance compared to individual optimal
controllers. The digital simulation results show the improved performance compared to
individual LQR controllers.

Chapter 8
CONCLUSION AND FUTURE WORK

8 CONCLUSION AND FUTURE WORK
8.1 Conclusion
Developing a hybrid model for the synchronous machine infinite bus (SMIB) power system
with the switching operation of FACTS devices to damp the power system oscillations was
the set main objective of the research. The task was successfully achieved by introducing
novel switching techniques in various combinations for PSS & UPFC based optimal control
inputs in power system to enhance the system dynamic performance and to damp the
power system oscillations.
The research work started with introduction to power system, classification of

stability and the overview of SMIB based Phillips heffron model. The review of PSS
design has been analyzed & it reveals that for the heavy complex networks under large
disturbances the PSS may unable to damp oscillations which may in turn leads to system
instability. To overcome this situation additional damping device Flexible AC Transmis-
sion Systems (FACTS) are used for power system networks. The comparison of various
FACTS devices are reviewed from earlier efforts and finally, considered the UPFC based
FACTS controllers, which was the most promising device in FACTS family for the current
research.
The design of optimal controllers LQR & LQG is discussed in this thesis. For, the
current research LQR control is designed for the power system models when the system is
not subjected to disturbance, when the system is under disturbance the LQG controllers
are proposed. Introduction to switched linear systems, motivation for switching in LTI
systems are also detailed in this thesis. The two standard switching control algorithms
which were applied for the power system models with modifications based on switching
requirements are explained along with the stability proof. Optimization via switching tech-
niques is also derived in this thesis to show the effectiveness of switching in optimization
of output energy compared to without switching.

With these concepts of power system & switched linear control theory, the switch-
ing strategies are proposed for the PSS & FACTS based optimal controllers in various
combinations to fulfill the objectives of the research. The remaining part of this chapter
will highlight the important results obtained in the thesis as contributions and introduces
new directions for future research.
8.2 Contributions of the thesis
The contributions of the thesis can be summarized as follows:
• For, the linearized Phillips heffron SMIB PSS model, the two feedback controllers
are designed using existing LQR & pole placement technique. The controllers are
designed in such a way that it must satisfy the requirements of two switching con-
trol algorithms. Switching between these two feedback controllers are experimented
with both modified switching control algorithms and the results are compared. Later,
switching concept is extended to UPFC FACTS device. Switching between two op-
timized feedback controllers are proposed using Lyapunov based modified switching
control algorithm for all the UPFC control inputs mE , δE , mB & δB . The proposed
switching model is simulated using simulink and the results are compared to select
the best UPFC control input.
• To, validate the effectiveness of coordinated design of PSS & UPFC compared to un-
coordinated design of PSS & UPFC an initial preliminary analysis is demonstrated.
Later, to improve the performance of the system with respect to peak overshoots &
settling time an switching between optimized & non optimized feedback controllers
are proposed for the simultaneous coordinated design of PSS+UPFC.
• From, the theory of switched linear control theory it reveals that intelligently, switch-
ing between two subsystems combines the useful properties of both. In the context
of UPFC based power system community earlier efforts are selecting the best UPFC

control inputs by applying different control techniques. Instead of selecting one

UPFC control input, this research combines the properties of two UPFC control
inputs by intelligently switching between two UPFC control inputs in various com-
binations and also with & without the coordinated PSS control input. In, the present
work power system is also subjected to disturbance and solved by using LQG. Switch-
ing between two LQG feedback controllers are also presented for all the UPFC control
inputs and the results are compared.
• Due to huge variations in the load power system installed with UPFC always oper-
ates at multi operating conditions. In this regard, one optimal controller for all the
loading conditions is quite a challenge. To, solve this problem in the present work
three different controllers are optimized for the best performance at each operating
(loading) conditions and switching concepts are introduced to switch between two
operating conditions based on three scenarios to select appropriate feedback con-
troller to take place which was optimized for the particular operating conditions to
get an overall controller with better performance at all operating conditions.
• In the present work, the preliminary analysis is developed for UPFC by designing the
feedback control signals using LQR based optimal control by tuning the weighting
matrices Q & R with different approaches. The approaches for the current research
are Bryson rule, Boudarel rule & multistage LQR. The designed three approaches
of LQR are applied for three operating conditions (light, normal & heavy loads)
of power system installed with UPFC and the results are compared to select the
better optimized feedback controllers for the specified operating condition. Later,
switching control strategy is developed to switch between master and alternative
controller (better selected optimal controllers from preliminary analysis) to show
further performance in damping of low frequency oscillations as well as optimization.

8.3 Recommendations for future work
This research has explored some good switching techniques for the PSS & UPFC (FACTS
device) for damping oscillations in power system. Based on the experience gained from
this research work, the following aspects are recommending for future work:
8.3.1 Switching between feedback controllers
In chapter 3, switching between two optimal state feedback controllers are proposed for
PSS as well as UPFC based power system for damping power oscillations. Switching
between two optimal output feedback controllers for power system are recommending
for future investigations to compare with the proposed state feedback switching control
research in chapter 3.
8.3.2 Switching control for Coordinated PSS & UPFC
Switching control is proposed for coordinated design of PSS & UPFC to damp power
oscillations in power system in chapter 4. The other advanced FACTS devices like IPFC,
STATCOM are also experimented with the coordinated design of PSS by implementing
with switching strategy remains as an open issue.
8.3.3 Switching strategies with disturbance
To, combine the properties of two UPFC control inputs, switching between UPFC controls
inputs using LQR state feedback control in various combinations is conducted in chapter
5. Switching between two LQG controllers with individual UPFC control inputs are also
suggested in this chapter when the system is subjected to disturbance. In future, recom-
mending carrying out the research to switch between different UPFC control inputs in
various combinations using LQG control with disturbances.

8.3.4 Optimal switching strategies: Loading conditions
In chapter 6, switching control strategy is developed to switch between two operating

conditions (light/normal, normal/heavy & light/heavy) with optimizing each individual
operating condition. This proposed switching techniques are depends on state of the
system. In future recommendations are suggesting designing a non dependent switch-
ing technique to switch between three operating conditions simultaneously, which needs
further analysis.
8.3.5 Multi LQRs Switching
In chapter 7, Multi LQRs switching is proposed for UPFC. In future recommends carrying
out the research work of switching between the various operating conditions as second
stage of switching where, the feedback controllers gains are taken from the first stage of
swicthing from multi LQRs. Combination of these two stages of switching is referred to as
multi stage switching control of multi LQRs for UPFC to cover all the loading conditions
in power system.
8.3.6 General Recommendations
The general recommendations other than chapter wise recommendations are:
1. To implement the various proposed novel switching techniques for multi machine
networks of UPFC & PSS in power system.
2. To implement other optimal feedback controllers like H-Infinity, Tabu seach algo-
rithm, etc., for the proposed novel switching techniques in future.

REFERENCES
[1] Douglas Leith, Robert Shorten, William Leithead and Oliver Mason, “Issues in the
design of switched linear control systems: A benchmark study,” International Journal
of Adaptive Control and Signal Processing, John Wiley & Sons Publishers, Vol. 17,
Issue 2, 2003, pp 1-5. DOI: 10.1002/acs.741.
[2] M. Sobha, R.Sreerama Kumar Saly George, “ANFIS Based UPFC Supplementary Con-
troller for Damping of Low Frequency Oscillations in Power Systems”, International
Journal of Electrical Systems, Vol.3, Issue.4, pp. 227-244, December 2007.
[3] E. V. Larsen, J. S. Gasca, and J. H. Chow, “Concepts for Design of FACTS Controllers
to Damp Power Swings”, IEEE Transactions on Power System, Vol. 10, No. 2, May
1995.
[4] S.Eshtehardiha and Gh.Shahgholian, “Improvement of STATCOM Per-

formance with Optimum LQR and Pole placement Controller Based
on Genetic Algorithm”, Proceedings of First Joint Conference on Fuzzy
and Intelligent Systems, Ferdowsi University of Mashhad, Iran, 2007.
http://confbank.um.ac.ir/modules/confdisplay/isfs2007/pdf/i637.pdf.
[5] X. P. Zhang, C. Rehtanz, B. Pal, “Flexible AC Transmission Systems: Modelling and

Control”, Springer, 2006, pp.317-380.
[6] Sidhartha Panda and Narayana Prasad Padhy, “ MATLAB/SIMULINK Based Model
of Single Machine Infinite-Bus with TCSC for Stability Studies and Tuning Employing
GA”, International Journal of Computer Science and Engineering, vol.1, pp.1-10.
[7] N. G. Hingorani, “High Power Electronics and Flexible AC Transmission System”,

IEEE Power Engineering Review, July 1988.

[8] N. G. Hingorani, “FACTS-Flexible AC Transmission System”, Proceedings of 5th Inter-

national Conference on AC and DC Power Transmission-IEE Conference Publication
345, 1991, pp. 1-7.
[9] N. G. Hingorani, “Flexible AC Transmission”, IEEE Spectrum, April 1993, pp.40-45.
[10] H. F. Wang and F. J. Swift, “A Unified Model for the Analysis of FACTS Devices
in Damping Power System Oscillations Part I: Single-machine Infinite-bus Power Sys-
tems”, IEEE Transactions on Power Delivery, Vol. 12, No. 2, 1997, pp. 941-946.
[11] M. A. Abido and Y. L. Abdel-Magid, “Power System Stability Enhancement via

coordinated design of PSS and FACTS-Based stabilizers”, Proceedings of 10th IEEE
International Conference on Electronics, Circuits and Systems, IEEE Xplore, 2003,
pp.850-853.
[12] M. A. Abido and Y. L. Abdel-Magid, “Analysis and Design of Power System Stabi-
lizers and FACTS Based Stabilizers Using Genetic Algorithms”, Proceedings of Power
System Computation Conference PSCC-2002, Session 14, Paper 3, Spain, and June
24-28, 2002.
[13] X. Chen, N. Pahalawaththa, U. Annakkage, and C. Kumble, “Controlled Series Com-

pensation for Improving the Stability of Multimachine Power Systems”, IEE Proceed-
ings - Generation, Transmission and Distribution, Vol. 142, Issue 4, IEEE Xplore,
1995, pp. 361-366.
[14] J. Chang and J. Chow, “Time Optimal Series Capacitor Control for Damping Inter-
Area Modes in Interconnected Power Systems”, IEEE Transactions on Power Systems,
Vol.12, No. 1, 1997, pp. 215-221.
[15] T. Lie, G. Shrestha, and A. Ghosh, ”Design and Application of Fuzzy Logic Control
Scheme for Transient Stability Enhancement in Power System”, Electric Power System
Research, Elsevier Journal, 1995, pp, 17-23.

[16] Y. Wang and G. Guo, “Robust nonlinear coordinated excitation and TCSC control
for power system”, IEEE Proceedings on Generation, Transmission and Distribution,
vol. 149, no,3, May 2002, pp. 367-372.
[17] M. A. Abido, “Genetic-based TCSC damping controller design for power system sta-
bility enhancement”, International Conference on Electric Power Engineering, Power
Tech Budapest 99, 1999, IEEE Xplore, 2002, pp. 165.
[18] M. A. Abido, “Pole placement technique for PSS and TCSC-based stabilizer design
using simulated annealing”, Electric Power System Research, Elsevier Journal, Vol.
22, 2000, pp. 543-554.
[19] Y. Wang, R. Mohler, R. Spee, and W. Mittelstadt, “Variable Structure FACTS Con-
trollers for Power System Transient Stability”, IEEE Transactions on Power Systems,
Vol. 7, 1992, pp. 307-313.
[20] T. Luor and Y. Hsu, “Design of an Output Feedback Variable Structure Thyristor
Controlled Series Compensator for Improving Power System Stability”, Electric Power
Systems Research, Elseveir Journal, 47, 1998, pp. 71-77.
[21] V. Rajkumar, R. Mohler, “Bilinear generalized predictive control using the thyristor
controlled series capacitor”, IEEE Transactions on Power Systems, Vol. 9, Issue 4,
1994, pp. 1987-1993.
[22] Q. Zhao and J. Jiang, “A TCSC Damping Controller Using Robust Control Theory”,
International Journal of Electrical Power and Energy Systems, Vol. 20, No. 1, 1998,
pp. 25-33.
[23] S.Eshtehardiha & Gh.Shahgholian, “Coordinating the Multivariable state fedback

controller on static synchronous compensator with Genetic Algorithm ”, Interna-
tional conference on intelligent and advanced systems, Ferdowsi University of Mashhad,
IEEE, PP 864-869, Aug 2007.

[24] Dr. Akram F. Bati, member IEEE, “Damping of Power Systems Oscillations by using
Genetic Algorithm-Based Optimal Controller ”, Iraq Jornal of Electrical and Electronic
Engineering, Vol.6, No.1, 2010, pp. 50-55.
[25] R. K. Pandey, Senior Member IEEE, “Analysis and Design of Multi-Stage LQR
UPFC”, of International Conference on Power, Control and Embedded Systems
(ICPCES), IEEE Xplore, 2010, pp. 4, DOI: 10.1109/ICPCES.2010.5698644.
[26] A. Safari , H. Shayeghi & H.A. Shayanfar, “A New Algorithm for Optimal Tuning of
FACTS Damping Controller”, International Journal on Technical and Physical Prob-
lems of Engineering (IJTPE), Issue 8, Volume 3 , Number 3, September 2011 , pp.
32-37, ISSN 2077-3528.
[27] Sasongko Pramono Hadi, “Dynamic Modeling and Damping Function of GUPFC in
Multi-Machine Power System”, IPTEK, The Journal for Technology and Science, Vol.
22, No. 4, November 2011, pp. 205-2013.
[28] Mehdi Nikzad, Shoorangiz Shams Shamsabad Farahani, Mehdi Ghasemi Naraghi,
Mohammad Bigdeli Tabar and Ali Javadian, “Comparison of robust control methods
performance in the UPFC controllers design”, Indian Journal of Science and Technol-
ogy, Vol. 4, No. 6, June 2011, ISSN: 0974- 6846, pp. 670-676.
[29] Doradla. Prathap Hari Krishna & M. Ravindra Babu, “Comparison of Different Tech-
niques for Design of Power System Stabilizer”, International Journal of Engineering
Science and Technology (IJEST), ISSN : 0975-5462, Vol. 3, No. 5, May 2011, pp.
4061-4066.
[30] Cuk Supriyadi Ali Nandar, “Design of Robust Power System Stabilizer Considering
Less Control Energy”, International Journal of Power Electronics and Drive System
(IJPEDS), Vol.2, No.1, March 2012, pp. 99-106, ISSN: 2088-8694.
[31] A. Venkateshwara Reddy, Gurunath Gurrala & M.Vijay Kumar, “Design of Pole
Placement Power System Stabilizers for Multi-Machine Systems without the External

System Information”, International Journal of Mathematics, Engineering and Tech-

nology, Vol. 5, No. 1, January 2013, pp. 42-57.
[32] Balwinder Singh Surjan, “Linearized Modeling of Single Machine Infinite Bus Power
System and Controllers for Small Signal Stability Investigation and Enhancement ”,
International Journal of Advanced Research in Computer Engineering and Technology
(IJARCET),Vol. 1, Issue 8, October 2012, ISSN: 2278 1323, pp. 21-28.
[33] Sangu Ravindra & DR.V.C.Veera Reddy, “Self Tuning Controllers for Damping Low
Frequency Oscillations”, International Journal of Engineering Science and Technology,
ISSN : 0975-5462, Vol. 4, No.09, September 2012, pp. 4160-4167.
[34] A.V.Sudhakara Reddy, M. Ramasekhara Reddy & Dr. M. Vijaya Kumar, “Stability
Improvement During Damping of Low Frequency Oscillations with Fuzzy Logic Con-
troller”, International Journal of Engineering Research and Applications (IJERA),
ISSN: 2248-9622, Vol. 2, Issue 5, September- October 2012, pp.1560-1565 .
[35] Ahad Jahandideh Shendi & Ali Ajami , “Application of UPFC Tuned Based on Grav-
itational Search Algorithm to Damping of Low Frequency Oscillations”, International
Journal of Advanced Research in Computer Science and Software Engineering, Vol. 3,
Issue 7, July 2013, ISSN: 2277 128X, pp. 762-771.
[36] Balwinder Singh Surjan, Ruchira Garg,“Power System Stabilizer Controller Design for
SMIB Stability Study,” International Journal of Engineering and Advanced Technology
(IJEAT), ISSN: 2249 8958, Vol. 2, Issue-1, October 2012, pp 209-214.
[37] Hisham M. Soliman, Ehab H. E. Bayoumi and Mohamed F. Hassan,“PSO based power
system stabilizer for minimal overshoot and control constraints,” Journal of Electrical
Engineering,Vol. 59, No. 3, 2008, pp 153159.
[38] Ali. M. Yousef and M K Ei-Sherbiny,“Improvement Of Synchronizing and Damping

Torque Coefficients Based LQR Power System Stabilizer,” Proceedings of International

Conference on Electrical, Electronic and Computer Engineering, IEEE Xplore, Sept.

2004, pp. 753 - 758, ISBN: 0-7803-8575-6, DOI:10.1109/ICEEC.2004.1374587.
[39] A. Alfi and M. Khosravi, “Optimal power system stabilizer design to reduce low
frequency oscillations VIA an improved system optmization algorithm”, International
Journal on Technical and Physical Problems of Engineering, Vol. 4 No. 2, Issue 11,
Jun 2012, pp 24-33.
[40] H.S. Ko, K.Y. Lee and H.C. Kim,“An intelligent based LQR controller design to
power system stabilization” Electric Power Systems Research, Elsevier Journal, Vol.
71, 2004, pp 1-9.
[41] Ali. M. Yousef and Ahmed M Khan ,“Opti- mal power system stabilizer based en-
hancement of synchronizing and damping torque coefficients”, WSEAS Transactions
on power systems, Vol. 7, Issue 2 ,ISSN :2224-350X, April 2012, pp 70-78.
[42] Sai Shankar, K T Veeramanju and Yathisha L, ”Application and Comparison of

Optimum Linear Quadratic Regulator Controllers for the Improvement of Static Syn-
chronous Compensator Performance”, Journal of Mechatronics, American Scientific
Publisher, Vol. 3, Issue 4, pp. 153-156.
[43] Ali. M. Yousef and Mohamed Zahran, “Improved Power System Stabilizer by Apply-
ing LQG Controller”Advances in Electrical and Computer Engineering, Vol. 8, ISBN:
978-1-61804-279-8, 2008, pp. 117-127.
[44] HaiFeng Wang, “A Unified Model for the Analysis of FACTS Devices in Damping
Power System OscillationsPart III: Unified Power Flow Controller”, IEEE Transactions
on Power Delivery, VOL. 15, NO. 3, July 2000, pp. 978-983.
[45] H. Shayeghi a, H.A. Shayanfar b, S. Jalilzadeh c and A. Safari, ‘Design of out-

put feedback UPFC controller for damping of electromechanical oscillations us-
ing PSO”, Energy Conversion and Management, Elsevier Journal, 2009, pp. 1-8,
DOI:10.1016/j.enconman.2009.06.005.

[46] N Tambey and M L Kotharir, “Unified Power Flow Controller (UPFC) Based Damp-
ing Controllers for Damping Low Frequency Oscillations in a Power System”, IE(I)
Journal-EL, Vol 84, June 2003, pp. 35-41.
[47] A.K. Baliarsingh, S. Panda, A.K. Mohanty and C. Ardil, “UPFC Supplementary
Controller Design Using Real-Coded Genetic Algorithm for Damping Low Frequency
Oscillations in Power Systems”, World Academy of Science, Engineering and Technol-
ogy, Vol 63, 2010, pp. 759-761.
[48] Vitthal Bandal, B. Bandyopadhyay and A. M. Kulkarni, “Output Feedback Fuzzy

Sliding Mode Control Technique Based Power System Stabilizer (PSS) For Single Ma-
chine Infinite Bus (SMIB) System” IEEE Transactions on Power Systems, 2005, pp
341-346.
[49] Gyugyi L “A Unified power-flow control concept for flexible ac transmission systems”,
IEE Proceedings on Generation, Transmission and Distribution Vol. 4, 1992, pp. 323-
331.
[50] Hingorani N. G., Gyugyi L “Understanding FACTS: concepts and technology of flex-
ible AC transmission systems”, Wiley-IEEE Press, 1999.
[51] Vilathgamuwa, M. Zhu and X.Choi “A robust control method to improve the per-
formance of a Unified Power Flow Controller”, Electrical Power Systems Research,
Elsevier Journal, Vol. 55, 2000, pp. 103-111.
[52] Amin Safari and Shayeghi,“Optimal Design of UPFC Based Damping Controller using
Iteration PSO,” International Journal of Electrical Power and Energy System Engi-
neering, pp. 151-156, 2009.
[53] Pourbeik P and Gibbard M. J. : Simultaneous coordination of power-system stabi-

lizers and FACTS device stabilizers in a multimachine power system for enhancing
dynamic performance, IEEE Transaction on Power Systems Vol 13, No. 2. 1998, pp.
473-479.

[54] Cai L. J. and Erlich I, “Simultaneous coordinated tuning of PSS and FACTS damping
controllers in large power systems, IEEE Transaction on Power Systems, Vol. 20, No.
1, 2005, pp. 294-300.
[55] Kwang M. Son and Jong K. Park, “On the Robust LQG Control of TCSC for Damping
Power System Oscillations”, IEEE Transaction on Power Systems, Vol. 15, NO. 4,
November 2000, pp. 1306-1312.
[56] Amir Elahi, Alireza Gholizadeh and Amin Aghae, “Linear Quadratic Gaussian Con-
trol for UPFC Auxilliary Stabilizer”, Sci.Int(Lahore) 2014, ISSN: 1013-5316, pp . 1535-
1538.
[57] M. Margaliot D. Liberzon, “Lie algebraic stability condition for nonlinear switched
systems and differential inclusions”, Systems and control letters, Elsevier Journal, pp.
1-13, 2001.
[58] J. P. Hespanha D. Liberzon and A. S. Morse, “Stability of switched system: a lie-

algebraic condition”, Systems and control letters, Elsevier Journal, pp. 117-122, 1999.
[59] S .S. Ge Zedong Sun, “Analysis and synthesis of linear control systems”, Automatica,
Elsevier Journal 41(2), pp. 181-195, February 2005.
[60] D. Z. Zheng Zedong Sun, “On reachability and stabilization of switched linear control
systems”, IEEE Transactions on Automatic Control, 46(2), pp. 291-295, 2001.
[61] Zedong Sun, “Canonical forms of switched linear control systems”, Proceedings of
IEEE American control conference, pp. 5182-5187, July 2004.
[62] A. Nerode P. J. Antsaklis, “Guest editorial hybrid control systems: an introductory

discussion to special issue”, IEEE Transactions on Automatic Control, pp.457-460,
April 1998.

[63] L. Magnani A. Ferrate and R. Scattlolini, “A globally stabilizing hybrid variable

structure control strategy”, IEEE Transations on Automatic Control, Vol. 47, Issue 8,
Aug 2002, pp.1334-1337.
[64] M. S. Barnicky, “Stability of switched and hybrid systems”, Proceedings of 33rd IEEE
Conference on Decision and Control, Dec 1994, pp. 3498-3503.
[65] Liberzon D and Morse A, “Basic problems in stability and design of switched sys-
tems”, IEEE Control System Magazine, Vol. 19, Issue 5, 1999, pp. 59-90.
[66] Lin H and Antsaklis P, “Switching stabilizability for continuous-time uncertain

switched linear systems”, IEEE Transactions on Automatic Control, 52(4):633-646,
2007.
[67] Lin H and Antsaklis P, “Stability and stabilizability of switched linear systems: a
survey of recent results”, IEEE Transactions on Automatic Control, Vol. 54, Issue 2,
2009, pp. 308-322.
[68] Liu X, “Stability analysis of switched positive systems: a switched linear copositive
lyapunov function method”, IEEE Transactions on Circuits and Systems, Vol. 56, Issue
5, 2009, pp. 414-418.
[69] Zhi Hong Huang, Cheng Xiang, Hai Lin and Tong Heng Lee, “A Necessary and
Sufficient Condition for Stability of Arbitrarily Switched Second-Order LTI System:
Marginally Stable Case” , 22nd IEEE International Symposium on Intelligent Control
Part of IEEE Multi-conference on Systems and Control Singapore, 1-3 October 2007,
pp. 83-88.
[70] Keith R. Santarelli and Munther A. Dahleh, ‘Comparison between a switching con-
troller and two LTI controllers for a class of LTI plants”,International Journal on
Robust Nonlinear Control, 2008, pp. 1-33, DOI: 10.1002/rnc.1308.

[71] Keith R. Santarelli and Munther A. Dahleh, “L2 Gain Stability of Switched Output
Feedback Controllers for a Class of LTI Systems”, IEEE Transactions on Automatic
Control, VOL. 54, NO. 7, July 2009, pp. 1504-1514.
[72] Keith R. Santarelli and Munther A. Dahleh, “Optimal controller synthesis for a class
of LTI systems via switched feedback”, Systems and Control Letters, Elsevier Journal,
VOL. 59, Issue No. 3, March 2010, pp. 258-264.
[73] L. Aravena and Lalitha S Devarakonda, “Performance driven switching control,”

IEEE Transactions on Industrial Electronics, E-ISBN: 1-4244-0497-5, July 2006, PP
31-36. DOI:10.1109/ISIE.2006.295564.
[74] Keith R. Santarelli, “A Switched State Feedback Law for the Stabilization of LTI
Systems” IEEE Transactions on Automatic Control ,Volume:56, Issue: 5 ,ISSN :0018-
9286, May 2011, pp 998-1013.DOI:10.1109/TAC.2010.2073590.
[75] Lalitha S. Devarakonda “Performance Based Switching Control for Single Input LTI
Systems”, M.S. Thesis, Department of Electrical Engineering, Graduate Faculty of the
Louisiana State University and Agricultural and Mechanical College DEC 2005.
[76] Zhiming Fang ,Zhengrong Xiang,Qingwei Chen and Jing Hua, “Optimal Switching
Sequence of a Class of Switched Systems with Parameter Uncertainty”, International
Journal of Nonlinear Science, World Academic Press, World Academic Union, Jan
2010, Vol 9, ISSN:1749-3897, pp 233-240.
[77] Tuhin Dasa and Ranjan Mukherjeeb, “Optimally switched linear systems”, automat-
ica, Science Direct, Elsevier, Mar 2008, pp. 1437-1441.
[78] Keith R Santarelli and Munther A Dahleh“Optimal Controller Synthesis for a Class
of LTI Systems Via Switched Feedback”, Systems & Control Letters, elsevier Journal,
Science Direct, Vol. 59, March 2010, pp. 258-264, DOI; 10.1016/j.sysconle.2010.02.003.

[79] Shengxiang Jiang and Petros G. Voulgaris“Performance Optimization of Switched

Systems: A Model Matching Approach”, IEEE Transactions On Automatic Control,
Vol. 54, No. 9, September 2009, pp. 2058-2071.
[80] H. Kawashima, Y. Wardi, D. Taylor and M. Egerstedt “Switching Control in DC-

DC Converter Circuits: Optimizing Tracking-Energy Tradeoffs”, Regular Paperl, The
Graduate School of Informatics, Kyoto University, Kyoto, Japan.
[81] Hussain N, Al Duwaish and Zakariya M, “A neural network based adaptive sliding
mode controller: Application to a power system stabilizer,” Elsvier Proceedings, Energy
Conversion and Management, Volume 52, Issue 2, February 2011, pp 1533-1538.
[82] Germane Xavier Athanasius, Hemanshu R. Pota and Valery Ugrinovskii “Robust
Decentralized Switching Power System Stabilisers for Interconnected Power Grids:
Stability using Dwell Time,” Proceedings of the 17th World Congress, The Interna-
tional Federation of Automatic Control Seoul, Korea, July 6-11, 2008, pp 8419-8424,
DOI:10.3182/20080706-5-KR-1001.1551.
[83] Al. Olimat, K.S. Farhoud and Hurtig, J.K, “Power System Stabilizers with
Fuzzy Logic Switching”, Power Systems Conference and Exposition, 2006, PSCE
’06. 2006 IEEE PES, ISBN:1-4244-0177-1, 2006, pp 2152 - 2157 , DOI:
10.1109/PSCE.2006.296277.
[84] Vitthal Bandal, B. Bandyopadhyay and A. M. Kulkarni, “Output Feedback Fuzzy

Sliding Mode Control Technique Based Power System Stabilizer (PSS) For Single Ma-
chine Infinite Bus (SMIB) System” IEEE Trans. On power systems,2005, pp 341-346.
[85] N V Ramanai, “Power System Operation and Control” Pearson Education, Chennai
Publishers, ISBN:978-81-317-5591-4. pp. 82-90.
[86] Anuj Banshwar, “Power System Analysis” Technical Publications, VTU, ISBN:
9789350992869. pp. 6.1-6.17.

[87] Prabha Kundur, “Power System Stability and Control” Tata MCGraw-Hill Publica-
tions, EPRI, ISBN: 978-0-07-063515-9. pp. 700-710.
[88] Vireshkumar G mathad and Basangouda F Ronad, “Review on comparison of FACTS

controllers for power system stability enhancement” International Journal of scientific
and research publications, Vol. 3, Issue 3, March 2013, ISSN: 2250-3153, pp 1-4.
[89] Mchrdad Ahmadi Kamarpshti and Mostafa Alinezhad, “Comparison of SVC, STAT-
COM, TCSC and UPFC controllers for static voltage stability evaluated by continous
power flow method” IEEE Electrical power and Energy conference, 2008, ISSN: 978-1-
4244-2895.
[90] Brian D. O. Anderson and John B. Moore, “Optimal Control: Linear Quadratic
Methods” Dover Books of Engineering, 2007.
[91] Yathisha L and S Patil Kulkarni , “Optimum LQR Switching Approach for the Im-
provement of STATCOM Performance” Springer LNEE, Vol 150, E-ISSN: 1876-1100,
2013, pp. 259-266, DOI: 10.1007/978-1-4614-3363-7 28.
[92] Sigurd Skogestad and Ian Postlethwaite , “Multivariable Feedback Control Analysis
and Design” Second Edition,John Wiley and Sons Publishers, August 2001, pp 359-362.
[93] Xuejiao Yang and Ognjen Marjanovic , “LQG Control with Extended Kalman Fil-
ter for Power Systems with Unknown Time-Delays” 18th International Federation of
Automatic Control (IFAC),World Congress Milano (Italy) August 28 - September 2,
2011, pp. 3708-3713.
[94] Keith R. Santarelli, “A Switched State Feedback Law for the Stabilization of LTI
Systems” Discrete Math and Complex Systems, Sandia National Laboratories, 2009,
SAND2009-5902.
[95] Seyed Abbas Taher, Shahabeddin Akbari, Ali Abdolalipour and Reza Hematti, “De-
sign of Robust UPFC Controller Using H∞ Control Theory in Electric Power System”,
American Journal of Applied Sciences, ISSN 1546-9239, 2008, pp. 980-989.

[96] P W Sauer, “Power System Dynamics and Stability”, Prenice Hall, 1997.
[97] Ali Ajami and Hamed Asadzadeh, “ Damping of Power System Oscillations Using
UPFC Based Multipoint Tuning AIPSO-SA Algorithm”, Gazi University Journal of
Science, 2011, pp. 791-804.
[98] Amin Safari and Hossein Shayegi, “ Optimal Design of UPFC Based Damping Con-
troller Uing Iteration PSO”, International Journal of Electrical and Electronics Engi-
neering, Vol. 3, Issue 11, 2009, pp. 672-677.
[99] Sangu Ravindra, Dr. V. C. Veera Reddy and Dr. S. Sivanagaraju“ A UPFC damping
control scheme using Lead-Lag and ANN based Adaptive controllers”, International
Journal of Engineering Research and Technology (IJERT), Vol. 11, Issue 6, 2012, ISSN:
2278-0181, pp. 1-5.
[100] Ali Ajami and Reza GholIizadeh“ Optimal design of UPFC-based damping controller
using imperialist competitive algorithm”, Turk J Elec Engg. and Comp Sci, Vol. 20,
No. Sup.1, pp. 1109-1122. DOI:10.3906/elk-1102-1047.

APPENDIX

APPENDIX-A
Simulation Diagrams:
A1. Switching between two feedback controllers (K1 /K2 ) & Switching
between UPFC control inputs (B1 K1 /B2 K2 )

A2. Switching between two operating (loading) conditions with UPFC in

power system
Note: Where,
AL = System matrix of light load.
AH = System matrix of heavy load.
BL = Control input matrix of light load.
BH = Control input matrix of heavy load.
KL = Feedback controller gain of light load.
KH = Feedback controller gain of heavy load.
S= Switching supervisor matrix.
C=Observation matrix.

APPENDIX-B
Computational methods to determine the feedback controllers K
B1: Chapter 4
Feedback controllers design for PSS using MATLAB:
1. MATLAB Program for primary controller (K1 )
2. MATLAB Program for secondary controller (K2 )

Feedback controllers design for UPFC using MATLAB:
1. MATLAB Program for primary controller (K1 ) for mE
2. MATLAB Program for secondary controller (K2 ) for mE
Note 1: Similarly, for other UPFC control inputs δE , mB , δB the feedback con-
trollers gains K are calculated.
Note 2: Similarly, for loading conditions (light, normal & heavy) the weighting
matrices considered are Q = I & R=1 for evaluating the feedback controllers K.

Dissemination of the Work
Scopus Indexed Journals:
1. Yathisha L and S Patil Kulkarni, “LQR and LQG Based Optimal Switching Tech-
niques for PSS & UPFC in Power Systems”, paper is under second review for
Control Theory and Technology, Springer Journal. (SJR:Q2)
2. Yathisha L and S Patil Kulkarni, “Optimal Switched Feedback Controller Design

for the Simultaneous Coordinated Design of UPFC and PSS in Power System”,
Journal of Electrical Engineering, University of Polytechnia, Romania, Vol.16, Issue
4, November 2016, pp. 408-414. (SJR:Q3)
3. Yathisha L and S Patil Kulkarni, “Optimal switching strategy method Performance

in the design of UPFC controllers”, International Journal of control theory and
applications, International Science Press, Vol. 9, Issue 37, December 2016, pp. 909-
921. (SJR:Q1)
4. Yathisha L and S Patil Kulkarni, “Optimal LQR Switching Approach for the Im-
provement of STATCOM Performance”, Springer LNEE, Vol. 150, pp. 259-266,
Aug 2013. DOI: 10.1007/978-1-4614-3363-7 28. (SJR:Q3)
5. Yathisha L, Kourosh Davoodi and S Patil Kulkarni, “Performance Analysis of Switch-

ing Between Bryson,Boudarel & Multistage LQRs for Power System with UPFC at
Different Load Conditions”, paper is under review for Journal of IET Control Theory
and Applications, IEEE Transactions, UK. (SJR:Q1)
IEEE Proceedings: Indexed in Scopus
1. Yathisha L, Kourosh Davoodi and S Patil Kulkarni, “Optimal switching control

strategy for UPFC for wide range of operating conditions in power system”, 3rd
Indian Control Conference, Jan 4-6, 2017, IEEE Xplore, IIT Guwahati. pp. 225-
232. DOI: 10.1109/INDIANCC.2017.7846479.

2. Yathisha L and S Patil Kulkarni “Application and comparison of switching control

algorithms for power system stabilizer”, IEEE International Conference on Industrial
Instrumentation and Control (ICIC), IEEE Xplore, May 2015, Pune, pp 1300-1305.
DOI:10.1109/IIC.2015. 7150949.
Peer Reviewed International Journals:
1. Yathisha L and S Patil Kulkarni, “Optimal Feed-Back Switching Control for the
UPFC based Damping Controllers”, International Journal on Control System and
Instrumentation (IJCSI), Vol. 3, No.2, ISSN No: 2158-0006, March 2012, pp. 49-53.
DOI: 01.IJCSI.3.2.79.
International Conferences:
1. Yathisha L and S Patil Kulkarni, “Optimal Feed-Back Switching Control for the
UPFC based Damping Controllers”, Proceedings of the Second International Joint
Conference on Advances in Engineering Technology - AET 2011, pp. 145-149. DOI:
02.AEE.2011.02.79.
2. Yathisha L, S. Patil kulkarni and R.S. Ananda Murthy “Hybrid Modelling and
Switching Algorithm for Power System with FACTS based Controllers”, in the pro-
ceedings of the International Conference on System Dynamics and Control (ICSDC-
2010), Manipal on 19th -22nd, pp-367-372, August 2010.

Finalrep 1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Finalrep 1

Uploaded by

Copyright:

Available Formats

Optimal Switching Strategy for Power System with FACTS

Visvesvaraya Technological University, Belagavi

For the degree of

Dr. Sudarshan S Patil Kulkarni

Department of Electronics & Communication Engineering,

My Parents, Wife, Family Members

The candidate has fulfilled all the prescribed requirements.

Dr. Sudarshan S Patil Kulkarni

I am grateful to Dr. T Nagabushan, Principal and Dr. M N Shanmukhaswamy,

I express my thankfulness to the faculty, co-scholars and staff members of Depart-

I am highly grateful to the authorities of the Visvesvaraya Technological University,

I am highly grateful and indebted to the Chairman, Secretary, Treasurer and

I express my sincere thanks to Dr.L Basavaraj, Principal and Dr. Mahesh P

I express my deep sense of gratitude and reverence to my beloved father Sri.

1.1 Literature Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.1.1 Power System with FACTS Based Controllers . . . . . . . . . . . . 4

1.1.2 Switched Linear Systems . . . . . . . . . . . . . . . . . . . . . . . . 9

1.1.3 Switching Control Strategies for Power System . . . . . . . . . . . . 11

1.1.4 Drawbacks of the Existing Control Techniques for Power System . . 11

1.2 Objectives of the Research . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1.3 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1.4 Scope, Limitations and Delimitation’s of the Research . . . . . . . . . . . . 14

1.5 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.2 Power System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.2.1 Linearization Concepts . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.2.2 Power System Stability . . . . . . . . . . . . . . . . . . . . . . . . . 22

2.2.4 Flexible AC Transmission Sytems (FACTS) . . . . . . . . . . . . . 29

3 OPTIMIZATION USING SWITCHING CONTROL 35

3.2 Optimal Control Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

3.2.1 Linear Quadratic Regulator Control . . . . . . . . . . . . . . . . . . 35

3.2.2 Linear Quadratic Guassian Control . . . . . . . . . . . . . . . . . . 38

3.3 Switched Linear Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

3.3.1 Motivation for switching . . . . . . . . . . . . . . . . . . . . . . . . 41

3.3.2 Switching control Algorithms . . . . . . . . . . . . . . . . . . . . . 42

3.4 Optimization of Performance Index Via Switching Control . . . . . . . . . 46

3.4.1 Optimization of JSwitch Using Switching Algorithm I . . . . . . . . 47

3.4.2 Optimization of JSwitch Using Switching Algorithm II . . . . . . . . 48

4 MODELING, OPTIMAL CONTROL DESIGN & SWITCHING CON-

4.2 Linearized Model of SMIB with PSS . . . . . . . . . . . . . . . . . . . . . 56

4.2.1 Experiments using Switching Control Algorithms . . . . . . . . . . 58

4.2.2 Result Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

4.3 Linearized model of SMIB with UPFC . . . . . . . . . . . . . . . . . . . . 64

4.3.2 Result Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

4.5 Preliminary Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

4.6 Experiments using switching control algorithm . . . . . . . . . . . . . . . . 79

5 LQR AND LQG BASED OPTIMAL SWITCHING TECHNIQUES FOR

5.2 Power system modelling with PSS & UPFC . . . . . . . . . . . . . . . . . 87

5.3 Switching Control Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

5.3.1 Switching Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . 88

5.4 Experimental Set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

5.5.1 Scenario I: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

5.5.2 Scenario II: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

5.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

6 SWITCHING CONTROL TECHNIQUES VIA OPTIMIZATION FOR

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

6.3 Proposed switching control for different operating conditions . . . . . . . . 112

6.3.1 Switched Linear Control Systems . . . . . . . . . . . . . . . . . . . 112

6.3.2 Switching Control Techniques . . . . . . . . . . . . . . . . . . . . . 113

6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132

7 SWITCHING CONTROL OF MULTI- LQRs FOR UPFC AS APPLIED

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

7.2 Preliminary Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

7.2.1 Bryson Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137