Professional Documents
Culture Documents
DECLARATION
It is hereby declared that the work presented in this thesis is the outcomes of the investigation
performed by us under the supervision of Md. Shafayet Hossain, Lecturer, Depertment of Electrical
and Electronic Engineering, Faculty of Engineering, Prime University. We also declared that on part
of this project has been submitted elsewhere for the award of any degree.
SUPERVISEDBY
PREPAREDBY
______________
________________________
Jahangir Alam
ID
: 103030301004
Batch : 25th (EEE)
_________________
Page I
Acknowledgement:
Our deepest gratitude is to our advisor Md. Shafayet Hossain. His efficient analysis of technical
problems and solutions to them will always be a trait. We will always be indebted to our supervisor
for his unfailing support for us since our first day at the thesis period. We would like to thank Md.
Shafayet Hossain who in many ways has influenced this work and us through his professional advice.
We have spent many hours in our university for our thesis and we would like to thank to the officials
who helped us in many ways. We would like to thank Md. Mostak Ahmed, the Chairman of dept.
EEE of Prime University to support us and inspires us in many ways. We also would like to thank
Mafin Muntasir Rhman sir to encourage us & make inquire about our thesis.
Page II
Abstract:
In systems and control engineering, and in real practical situation suggests how the existing theory
needs to be modified to accommodate more realistic conditions on real plants. Much research has
been carried out in this area of designing controllers for real plants. Controller design is made easier
by the availability of information on the real plant (model structure, values of the plant parameters
and acting disturbances are well known); so it is straight forward to obtain a desired behavior of the
plant output. With earlier Adaptive Laws when the plant model structure is known but its parameters
are unknown, the process of developing a control scheme naturally becomes complicated. In general,
it has been a common experience to find that more complex is the plant model: the more complex is
the design of a suitable controller scheme. Adapting/adjusting the controller parameters for such
cases (where plant parameters were not accurately known), is known as Adaptive Control. A typical
adaptive controller scheme has mechanism in an additional loop (over and above the ordinary
feedback loop) for adjusting / adapting the controller parameters. The equations which describe how
these controller parameters need to be adjusted are called Adaptive Laws. Since the overall system
stability is the primary requirement, adaptive control schemes and adaptive laws have been developed
based on Lyapunov stability theory, historically since the 1960s.
Page III
Dedicated to
Our Parents
Honorable Supervisor, Md. Shafayet Hossain
and all the faculty members of
Electrical & Electronic Engineering Department
Page IV
Content
I
II
III
IV
V
VII
VII
Declaration
Acknowledgement
Dedication
Abstract
Content
List of Figures
List of Tables
Chapter-1(Introduction)
1.1
Background
1.2
Objectives
1.3
Contributions
Introduction
2.2
Related Works
2.3
2.4
2.4.1
Motivation of MRAC
2.5
2.6
2.7
2.8
2.9
Page V
Time Response
3..1.1
3.1.2
Time Constant
3.1.3
10
3.2
Working Principle
11
3.3
Components
11
3.4
11
3.5
Mathematical modeling
12
3.6
12
3.7
Loss Function
13
3.8
13
3.9
14
3.10
14
3.11
15
Page VI
16
17
19
21
Conclusion
Future works
23
24
Reference
26
List of figures
Figure-2.1
Figure-2.2
Figure-2.3
Direct MRAC
Figure-2.4
Indirect MRAC
Figure-2.5
Figure-3.1
Figure-3.2
10
Figure-4.1
17
Figure-4.2
18
Figure-4.3
19
Figure-4.4
20
Figure-4.5
PID Controllers
21
Figure-4.6
22
Figure-5.1
24
Figure-5.2
25
List of Table
Figure-2.1
Page
VII
Chapter 1: Introduction
1.1 Background
This thesis presents the elements of model reference adaptive control, which refers to a particular
control procedure for uncertain dynamic systems. The control problem as well as the adaptive
control problem is described. The use of various models for control including identification model
and reference model is presented. The model-following in the presence adaptive control is
described. The error model approach for designing the requisite controllers is delineated. The
solution to the model reference adaptive control for linear plants is presented. Its extension to
nonlinear system is briefly mentioned. The role of parameter identification and its relation to
persistent excitation is described. Major developments in the field of model reference adaptive
control have taken place in the eighties and nineties and have been applied in a number of
practical control problems with success. Adaptive control is a very appealing technology, at least
in principle. Yet its use has been conditioned by an attitude of distrustfulness on the part of some
practitioners. In this thesis, we explain why such distrustfulness is warranted, by reviewing a
number of adaptive control approaches which have proved deficient for some reason that has not
been immediately apparent. The explanation of the deficiencies, which normally were reflected in
unexpected instabilities, is our main concern. Such explanations, coupled with remedies for
avoiding the deficiencies, are necessary to engender confidence in the technology. These include
the unpredictable failure of the MIT rule; the bursting phenomenon, and how to prevent it; the
Rohrs counterexample, which attempted to disqualify all adaptive control algorithms; the notion
that identification of a plant is only valid conceptually for a restricted range of controllers (with
the implication that in adaptive control, certain controller changes suggested by adaptive control
algorithms may introduce instability); and the concept of multiple model adaptive control. The
adaptation law attempts to find a set of parameters that minimize the error between the plant and
the model outputs. To do this, the parameters of the controller are incrementally adjusted until the
error has reduced to zero. A number of adaptation laws have been developed to date. The two
main types are the gradient and the Lyapunov approach and we have use lyapunov approach.
1.2 Objectives
(1) Specify the desired control loop performances.
(2) To know the dynamic model of the plant to be controlled.
(3) To possess a suitable controller design method making it possible to achieve the
desired performance for the corresponding plant model.
(4) To minimize the error(ideally zero)
(5) to achieve and to maintain acceptable level of performance when plant (disturbance)
model parameters are unknown or vary
Page 1
1.3 Contributions:
The main contribution of this thesis is the design of final controller dynamics using MIT rule for
systems affected by both matched and mismatched types of uncertainty. The control law is
designed in such a way that the discontinuous sign function acts on the time derivative of the
control input. So the actual control obtained after integration is continuous and hence chattering is
eliminated. Adjustable control mechanism is used to eliminate the error. The proposed idea of
model reference adaptive control is used to design for the plant system where the plant is
unknown or change with time. Adaptive controllers are a fact of life, and have been for some
decades. However, theory and practice have not always tracked one another. In this paper, we
examine several instances of such a mismatch.
These are:
The MIT rule, an intuitively based gradient descent algorithm that gave unpredictable
performance; satisfactory explanation of performance started to become possible in the 1980s.
Bursting, a phenomenon of temporary instability in adaptive control algorithm implementation
of a type observed in the 1970s; explanation and our understanding of avoidance mechanisms
only became possible in the 1980s.
The Rohrs counterexample, which argued that adaptive control laws existing at the time could
not be used with confidence in practical designs, because unmodeled dynamics in the plant could
be excited and yield an unstable control system.
Iterative controller re-design and identification, an intuitively appealing approach to updating
controllers that came to prominence in the 1980s and 1990s, and which can lead to unstable
performance. Explanation and an understanding of an avoidance mechanism came around 2000.
Multiple model adaptive control, another intuitively appealing approach to adaptive control with
the potential to include non-linear systems.
Page 2
Page 3
Page 4
Page 5
Page 6
Page 7
Page 8
Page 9
Page
10
3.3 Components
Reference Model: It is used to give an idyllic response of the adaptive control system to the
reference input.
Controller: It is usually described by a set of adjustable parameters. In this paper only one
parameter is used to describe the control law. The value of is primarily dependent on
adaptation gain.
Adjustment Mechanism: This component is used to alter the parameters of the controller so that
actual plant could track the reference model. Mathematical approaches like MIT rule, Lyapunov
theory and theory of augmented error can be used to develop the adjusting mechanism. In this
paper we are using MIT rule with Normalized Algorithm and the technique is then referred as
Modified MIT rule. The basic block diagram of MRAC system is shown in the fig.1. As shown in
the figure, ym(t) is the output of the reference model and y(t) is the output of the actual plant and
difference between them is denoted by e(t).
e(t) = y(t) - ym(t)
Page
11
k
y(t)=k.G(p) 0.uc(t) =k.G(p) ko.uc(t) = ym(t)
3.6 The MIT (Massachusetts Institute of Technology) Rule:
According to MIT rule, the cost function or loss functions defined as
F () = e2 ----------- (1)
Where, e is the difference of the output of the reference model and the actual model, while is the
adjustable parameter known as the control parameter. In this rule the parameter is adjusted in
such a way so that the loss function is minimized
d/dt = - F / ---- (2)
d/dt = - 2e e / --- (3)
e/ , is called the sensitivity derivative of the system. This shows how the error is dependent
on the adjustable parameter, . We have used the MIT rule defined by equations (1), (2) and (3)
for developing the control law.
Page
12
f (t,0) = e(t,0)
d
dt
t f +
f=
f .d 0
dt
2e
d
dt 0
.d 0
t e dt
2e
2e
d
dt 0
.d 0
t e dt
d
dt
= t k.G(p)(0(t).uc(t))
k.G(p)(uc(t))
d 0=
dt
where
2e
. ym(t).e(t,0)
0 is arbitrary since n =
Page
13
k
with arbitary
ko
d
dt
y= ay + bu,
y=
b
p+a u
d
y = am ym+ bmuc , ym=
dt m
bm
p+amuc
01= 01o =
am - a
bm ,
02 = 01o = b
b
b01
e(t)=y(t) - ym(t)=
p+a+b02
uc(t)-
bm
p+am
uc
Page
14
e
b
u (t)
=
01 p+a+b02 c
-b
b01
e
y(t)
= (p+a+b0
2 uc(t)=
p+a+b0
2
02
2)
If y(x)=u(x)/v(x)
Then dy/dx={v.(du/dx)-u.(dv/dx)}/v2
d
dt
01 = - .e(t)
d
dt
01 = -
d
dt
01 = - .e(t)
d
dt
01 =
n.e(t)
b
.e(t) . p+am uc(t)
e(t)
01
am u (t)
c
p+am
n.e(t)
-b
.e(t) . p+am uc(t)
e(t)
01
am u (t)
c
p+am
Where
b
am
Page
15
Page
16
4.2 Under-damped
Page
17
Page
18
4.3 Over-damped
Page
19
Page
20
Page
21
We observed from above simulation of MRAC &PID controller that PID controller varies if only
one parameter vary. But MRAC adjusts if plant parameter varies with time or unknown. This is
the main difference between them.
Page
22
+
C
V(t)
Vc(t)
I(t)
Summing the voltages around the loop, assuming zero initial conditions, yields the integral
differential equation for this network as
di(t ) + Ri(t ) + 1
C
dt
t
i ( ) d = v(t)
o
d2q(t)
1
dq(t)
q(t)= v(t)
+
+R
2
C
d(t)
dt
q(t) = Cvc(t)
Page
23
d2vc(t)
dvc(t)
LC
+ vc(t)= v(t)
+RC
2
d(t)
dt
Taking the Laplace transform assuming zero initial conditions, rearranging terms, and simplifying
yields
(LCs2+RCs+1)Vc(s)=V(s)
Solving for the transfer function, Vc(s) / V(s), we obtain
Vc(s)
=
V(s)
V(s)
1'LC
s2 +
R
1
s+
L
LC
1
LC
1
R
S2+
S+
L
LC
Vc(s)
R/L= 10
Page
24
References
Page
25