You are on page 1of 5

Fractional order controller to control pump in FESTO

MPS PA Compact Workstation

Andrzej Koszewnik Tomasz Nartowicz Ewa Pawuszewicz


Bialystok University of Technology, Bialystok University of Technology, Bialystok University of Technology,
Faculty of Mechanical Engineering Faculty of Mechanical Engineering Faculty of Mechanical Engineering
Wiejska 45C, Wiejska 45C, Wiejska 45C,
15-351 Bialystok, Poland 15-351 Bialystok, Poland 15-351 Bialystok, Poland
a.koszewnik@pb.edu.pl tomek.nartowicz@gmail.com e.pawluszewicz@pb.edu.pl

AbstractThe aim of the paper is to present the simple Testing Festo MPS PA Compact Workstation [16-18]
method of determining a fractional order controller to control causes practical training on industrial systems, project kits,
the pump in Festo MPS PA Compact Workstation. Transfer stations, systems and training factories, what is seldom
function of the controller follows directly from using of Bode's possible in reality.
ideal transfer function as a reference one for the open loop
The goal of the present study is to develop the simplest as
system. Such systems generally are unsensitive to gain changes in
the open loop. In Festo MPS PA Compact Workstation by using possible method of design of fractional order controller that
a corresponding controllers a level and a flow rate can be set up can be applied to control the pump in Festo MPS PA
as a cascade control system. The design process allows to use Compact Workstation. The transfer function of the pump was
conventional continuous P, I, PI, PID controllers and appointed by identification method [16-18].
discontinuous controllers (two-point controllers). In the To our aim the Bode's ideal transfer function is used as a
presented study the fractional order controller is tested to obtain reference transfer function for the open loop system. The idea
parameters controller of this system. of this approach was proposed in [19] in the classical case and
in [20] for the fractional case. Theoretical approach in the
present study is based on the approach proposed in [21] and
KeywordsFesto MPS PA Compact Workstation, fractional
order controller, Bode ideal transfer function, reference transfer
developed in [22-24] for a class of stable and unstable plants
function, phase margin with fractional order controllers. The form of the controller is
not impose, for every single investigated transfer function of
I. INTRODUCTION the object different form of the controller is given.
Recently the concept of fractional derivatives and II. THE LABORATORY STAND
differences is under strong consideration as a tool in
descriptions of behaviors of real systems. In control theory it The Festo MPS PA Compact Workstation shown in Fig.1
during last years one can observe a growing number of is a laboratory stand has been used to design fractional PID
applications of the fractional calculus, see for example [1-5]. controller. As we can see in the photo the stand composed
In natural way fractional order PID controllers, namely PID with a pump, many pipes, two tanks, other kinds sensors,
controllers, including an integrator of order and a actuators and PLC controller Siemens S7-300 allow to testing
of other types control systems.
differentiator of order, have received an attention in from
engineering point of view, see for example in [4-8]. In fact
such controllers are more robust to gain changes then classical
one. Since in a design process of fractional order controllers
usually five parameters should be to select, instead of three in
standard case, the tuning of the controller is more complicated.
Some design methods of tuning of PID controllers are
presented for example in [8-12].
One of the most developed approaches to design robust
and fractional order controllers is CRONE control
methodology, French acronym of Commande Robuste
dOrdre Non Entier (non-integer order robust control) [13-
15]. Fig.1 The photo of the laboratory stand Festo MPS PA Workstation
Fig.3. Scheme diagram of control system with negative feedback loop.
The Festo Didactic station allows to realize such control
system as: level, pressure, temperature and flow rate [25]. In where:
order to design FO controller the level control system has C(s) the transfer function of the fractional order controller
chosen that scheme is shown in Fig.2. G(s) - transfer function of the pump,
As it can be seen in Fig.2 process of level control system is yr(s) - input,
realized by the pump P101 pumps liquid from tank B101 to y(s) - output
tank B102 and the ultrasonic sensor located on the top of the
tank B102. A measurement signal in the control system is The pump as a control plant is a first order plant described
level of liquid in tank B102 but a actuator signal is a voltage in following form:
applied to the pump. Also, taking into account Fig.2 we seen kp
that the control system could be disturbance by manually G ( s) = , (1)
1 + s
open/close the value 102 or the value 110.
where:
kp the gain of the pump,
- the time constant (positive real number).

Transfer function of the proposed controller follows


directly from a using of Bode's ideal transfer function. In this
case such the transfer function is described by:


K (s) = c , (2)
s

where c is the gain crossover frequency ( | K ( jc ) |= 1 ) and


is a fractional order, is a reference transfer function of the
open loop system.

Transfer function of the open-loop system expressed in (2)


describes the derivative plant for < 0 and the integral plant
Fig.2. Scheme of level control system [28]. for > 0 . The control system described by (2) has constant
value of phase margin m = (1 0.5 ) , see [19-20]. Hence,
Taking into account other control systems which can be tested
on this stand it can be seen that the pump is a common device such system is insensitive to gain changes in the open loop
of all aforementioned systems. So, in order to ensure correctly system, see Fig.4. For detailed analysis of the presented
works of all control system the mathematical model for the system in Fig.3, including time domain, see [4].
pump should be determined. With help of Matlab software
such model is obtained and described in details in [26]. 20 log | K ( j ) |
In result of experimental identification procedure the obtained
model allow to design fractional order PID controller that 20 log | K ( j ) |
process is described in next sections. 0

log
III. PROBLEM FORMULATION OF DESIGN FRACTIONAL ORDER arg K ( j ) j c
CONTOLLER



The process design of fractional order controller has been 2
realized with using the scheme diagram of control system m = (1 / 2 )
shown in Fig.3. log

Fig.4 Logarithmic characteristics of system described by (2) for 1 < < 2


yr(s y(s)
- - C(s) G(s)
The reference transfer function (1) is used for the open loop
system. In effect, to obtain the ideal Bode transfer function in
open loop system, the controller transfer function should have
a structure:
1 + sT relations between gain margins and real order for the phase
C ( s) = k c
, (3) margin from 45 to even 80.
s
where is a real, not necessary integer, number. Assuming
that T = , the transfer function for the open loop system phase margin fractional
with the process (1) and with the fractional order controller (3) m order
is given by:
45 o 1,50
k p kc
K ( s ) = G ( s )C ( s ) = (4) 60 o 1,33
s
In order to analyze the process of design fractional order 80 o 1,11
controller for the transfer function (4) the control parameter
is searched for a fixed phase margin m of the open loop In result of the computer simulation the step responses of
system. Note that (4) has a form of considered Bode's ideal the control system for different values of phase margin m are
transfer function (2), so the constant value of phase margin is: obtained. As one can see in Fig.5 shorter settling time and
bigger overshoot are results of increase value of the phase
m = (1 0.5 ) (5) margin.

1.4
Next, taking into account (4) the real value is calculated and
expressed by:
1.2
2
= 2 m (6)

step response y(t) 1

Note once again that such system is unsensitive to gain


0.8
changes in open loop system and the fractional order of the
controllers is relative with phase margin only.
0.6
IV. RESULTS OF COMPUTER SIMULATION
The investigation of control system of the pump in Festo 0.4
MPS PA Compact Workstation with using fractional order
controller is carried in Matlab/Simulink software. The transfer 0.2 phase margin=45 deg
function of this pump obtained in identification procedure [27] phase margin=60 deg
is expressed by phase margin=80 deg
0
0 1 2 3 4 5 6
time [sec]
0,417
G( s) = , (7) Fig.5. Step responses of control system (Fig.3) with transfer function (7) and
1 + 0,5s
controller (3) for different values of phase margin m
for the fractional order controllers given by (3) for different
parameters are investigated. Next, the parameters of the closed-loop systems are
In results of the computer simulation the step responses of analyzed for the controller with order =1.11 and different
the consider closed-loop systems with different parameters of value of the gain of the open-loop system K = k p k c . Taking
the fractional order controller are obtained. The control systems
into account the Fig.6 it can be notice that the system is robust
are tested in two stages. In the first step the closed-loop system
to change of the parameter K in range of 5 to 200. Similar to
is investigated in depends on phase margin of the system.
previous obtained results also in this case shorter settling time
However in second step in depends on different value of gain
and bigger overshoot are obtained in result of increase of
K=kpkc where kp is the gain of the pump and kc is the gain of
parameter K. Taking into account the obtained results we can
the controller. Results of these investigations are shown in
suppose that further increase of the parameter K can only leads
Fig.4 and Fig.5, respectively.
to achieve more overshoot in very short time. Hence, in
Taking into account control theory of mechanical systems
control strategy point of view such system the best parameters
we known, that important feature of all closed loop systems is
of the closed-loop are obtained for the gain equals 100.
stability. In classical control systems the stability of the system
depends on phase margin that value is between 30 and 60.
For such values of phase margin the order of fractional order
controller was calculated based on equation (6). The obtained
order of the controller are collected in Tab1. Analysis of Tab.1
shows that there exists a strong dependence between the gain
margin and real orders of controller. Tab. 1 confirms such
1.2 REFERENCES
[1] S. Das, Functional Fractional Calculus for System Identification and
Controls, Springer, Berlin, 2008.
1 [2] T. Kaczorek, Selected Problems of Fractional Systems Theory,
Springer, Berlin, 2011.
[3] A. A. Kilbas, H. M. Srivastava, J. J. Trujillo, Theory and Applications of
0.8 Fractional Differential Equations, Elsevier, Amsterdam, 2006
step response y(t)

[4] P. Ostalczyk, Epitome of the Fractional Calculus, Theory and its


Applications in Automatics, Publishing Department of Technical
0.6 University of d, 2008, (in Polish).
[5] I. Podlubny, Fractional-order systems and PID-controllers, IEEE Trans.
on Automatic Control, Vol. 44, No.1, pp. 208-214, 1999
0.4
[6] D. Valerio, Fractional Robust Systems Control. PhD Dissertation,
Technical University of Lisbona, 2005.
K=5 [7] I. Podlubny, Fractional Differential Equations, Academic Press, San
0.2 K=10 Diego, 1999.
K=100 [8] R. Caponetto, G. Dongola, I. Fortuna, I. Petras, "Fractional Order
K=200 Systems. Modeling and Control Applications, World Scientific Series on
0 Nonlinear Science, Series A, Vol. 72, World Scientific Publishing, 2010.
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
time [sec] [9] J. A. T. Machado, M. F. Silva, R. S. Barbosa, I. S. Jesus, C. M. Reis, M.
G. Marcos, A. F. Galhano, "Some applications of fractional calculus in
Fig. 6. Step responses of control system (Fig.3) with transfer function (7) and
engineering", Mathematical Problems in Engineering, ID 639801, 2010.
controller (3) for different values of K
[10] A.. Monje, B. M. Vinagre, V. Feliu, Y. Chen, "Tuning and auto-tuning of
fractional order controllers for industry applications", Control
V. SUMMARY AND CONCLUSIONS Engineering Practice, Vol. 16, pp. 798-812, 2008.
[11] C. Monje, Y. Chen, B. Vinagre, D. Xue, V. Feliu, "Fractional-order
The problem of stability of control system composed with Systems and Controls", Spronger-Verlag, London, 2010.
first order plant and fractional order controller is presented in [12] D. Xue, Ch. Zhao, Y. Q. Chen, "Fractional order PID control of a DC-
motor with elastic shaft: a case study", Proc. American Control
the paper. Based on computer simulations the parameters of Conference, Minneapolis, USA, pp. 3182-3187, 2006.
the fractional controller are determined in results of using once [13] A. Oustaloup, La commande CRONE, Editions Hermes, Paris, 1991.
of simply method based on Bodes plot of reference transfer [14] A. Oustaloup, La derivation non entiere theorie, syntheses at
function open-loop system K(s). The obtained results. shown applications, Editions Hermes, Paris, 1995.
that the fractional order controller is related only to phase [15] A. Oustaloup, Diversisity and Non-iteger Differentiation for systems
margin but is not related with parameters of the plant. dynamics, Wiley 2014.
Confirmation of obtained results are presented in Fig.5 where [16] J. Helmich, Festo MPS PA Compact Workstation Manual, 2008.
the parameters of the control system are changed in spite of [17] E. Wolfgang, MPS PA Compact Workstation, Fluid Lab - PA -
constant value of controllers fractional order. If the gain of Handling description, 2008.
the fractional order controller is increasing then the overshoot [18] Festo Didatic Technical Documentation Data Sheet, 2008.
also is increasing but settling time is decreasing. [19] W. Bode, "Network Analysis and Feedback Amplifier", Van Nostrand,
New York 1945.
In order to validate the obtained results the parameters of
[20] Boudjehem, D. Boudjehem, H. Tebbikh, Simple analytical design
designed fractional controller will implemented to real-time method for fractional-order controller, Proc. 3-rd IFAC Workshop on
processor. Then, once again the open-loop system and the Fractional Differentiation and its Applications, Ankara, Turkey, (CD-
closed-loop system will investigated. Experimental results ROM), 2008.
allow to prove results obtained in computer simulations. [21] S. Barbosa, J. A. Machado, I. M. Ferreira, Tuning of PID controllers
based on Bode's ideal transfer function, Nonliner Dynamics, Vol. 38, pp.
305-321, 2004.
[22] M. Busowicz, T. Nartowicz, "The controller design for a unstable plant
with delay". Pomiary Automatyka Robotyka, Vol. 2, pp. 398-405, 2009,
ACKNOWLEDGMENT (In Polish
This work of A. Koszewnik and E. Pawluszewicz is financed [23] T. Nartowicz, " Design of a fractional order controller satysfying gain and
from by Bialystok University of Technology grant phase margin for a class of internal plants with delay", Pomiary
Automatyka Robotyka Vol. 56(5), pp. 409-413, 2010, (In Polish).
S/WM/1/2016.
[24] T. Nartowicz, "Design of Fractional Order Controller Satisfying Given
The work of T. Nartowicz is financed by Bialystok University Gain and Phase Margin for a Class of Unstable Plant with Delay", Acta
of Technology by his research university grant. Mechanica et Automatica 6, Vol. 6, No. 4, pp. 41-45, 2012.
[25] MPS-PA Collection of Datasheets_en_11-05-19.pdf
[26] P. Romuszewski, A. Mystkowski, E. Pawuszewicz, Modeling and
identification Festo MPS PA Compact Workstation, Przeglad
Elektrotechniczny, Vol. R. 91, nr 3, pp. 152-156, 2015.
[27] A.Koszewnik, Z.Gosiewski, Frequency Domian Identification of the
Active 3D Mechanical Structure for the Vibration Control System,
Journal of Vibroengineering, 2012, Vol. 14, No.2, pp.451-457.
[28] Festo Didactic Technical Documentation Data Sheet, (2008).

You might also like