Sei sulla pagina 1di 181

ECE 5470 Control Systems 2

ECE 5470 Control Systems 2


Introduction: instructor
Introduction: students
Reason(s) to take this class
Control System: Theory and Skills
Example: System Engineering V Model
Example: US 8706333
Reason(s) to take this class
Reason(s) to take this class
Adobe Acrobat
Document
Model: linear vs nonlinear
(eq. 10-11)
Observer ( piece-wise linear)
(eq. 12-13)
Stability (see above)
Adaptive observer
Kalman filter (parameter
identifier) (eq. 18)
Simulation (figure 6)
Topics
1. Analysis of Linear Systems (2.5 weeks)
Quiz at Week 4
2. Control of Linear Systems (2.5 weeks)
Quiz at Week 6
3. Stability Theory (2 weeks)
Quiz at Week 8
Project start at Week 7 or 8
4. Optimal Control (2.5 weeks)
5. Robust Control (1.5 week)
Expectations
1. State space approach to system modeling,
analysis and control design:
State Space modeling
Controllability/Observability
State Feedback control (with observer)
Stability
Optimal control
Robust control design
2. Tools associated with modeling and control
design (Matlab/Simulink/Stateflow)
HW/Quiz/Final/Project/Grade
No late submission of HW
No make-up quiz
No make-up final unless allowed by governing
rules
No late submission of project
Grading strictly follows that given in Syllabus
A
-
, B
+
, B
-
, C
+
, C
-
are consistent with the ranges set
in Syllabus
Class Arrangement & Office Hour
Class:
50 minutes lecture + 10 minutes break
Open to other suggestions
Office Hour:
12-1 pm same day/location as the class
Open to email correspondence
AN3400@WAYNE.EDU
Open to other suggestions
ECE 4330
ECE 4470
ECE 5440
ECE 5470
ECE 7420
ECE 7430 ECE 7440
ECE 7450
Part 1
Analysis of Linear Systems
Introduction
Mathematical Model
State Space Representation
Transfer Function
Responses of Linear Systems
Similarity Transformation
What about Federal Tax System? Is it linear or nonlinear?
Key to determine if a system is linear or nonlinear:
A linear system satisfies the Superposition Principle:
y
1
=S(u
1
) ^ y
2
=S(u
2
) ^ (a
1
, a
2
constant)
S(a
1
*u
1
+a
2
*u
2
)=a
1
*S(u
1
)+a
2
*S(u
2
)
Superposition principle includes two properties:
Additivity: S(u
1
+u
2
)=S(u
1
)+S(u
2
)
Homegenuity: S(a
1
*u
1
)=a
1
*S(u
1
)
Example: Linear Operator
Let D[x]=dx/dt, let L=D
2
+p*D+q*I be the operator
operating on related variables. Then if p, q are constants, it
is linear.
Proof: L is linear if for ANY constants a
1
, a
2
, and ANY
variables x
1
, x
2
, we have
L(a
1
*x
1
+a
2
*x
2
)=a
1
*L(x
1
)+a
2
*L(x
2
)
We need to show the above equality holds:
L(a
1
*x
1
+a
2
*x
2
)
= (a
1
*x
1
+a
2
*x
2
)+p*(a
1
*x
1
+a
2
*x
2
)+q*(a
1
*x
1
+a
2
*x
2
)
=a
1
*x
1
+a
2
*x
2
+p*a
1
*x
1
+p*a
2
*x
2
+q*a
1
*x
1
+q*a
2
*x
2
= a
1
*(x
1
+p*x
1
+q*x
1
)+a
2
*(x
2
+p*x
2
+q*x
2
)
= a
1
*L(x
1
)+a
2
*L(x
2
)
Example: Fuel Economy
Long term (new car vs old car)
Seasonal (Spring vs Winter)
Example: Battery Power Capability
A battery is a device that stores energy and then
provides energy to load
Same input (State of Charge) but at begin of life
(BOL) and end of life
power capabilities will be different
We studied the following system properties:
Linear vs Nonlinear
Time-invariant vs Time-Varying
Single Input Single Output (SISO) vs Multi- Input
Multi- Output (MIMO)
Continuous time vs Discrete Time
Examples of different types of controls:
Closed loop:
1. Air conditioning
2. Engine air/fuel ratio control
3. Cruise control (speed control)
4. *Battery state of charge estimation (voltage
feedback)
Open loop:
1. Toy motors
2. *Battery state of charge calculation (ampere-hour
integration)
We studied the following control types:
Open Loop (no-feedback)
Closed Loop (feedback)
Better Performance (including stability)
Cost more
What happens if angular
speed is measured as
output?
Example: DC Motor Model using
m-script
J = 3.2284E-6
b = 3.5077E-6
K = 0.0274
R = 4
L = 2.75E-6
A = [0 1 0
0 -b/J K/J
0 -K/L -R/L]
B = [0 0; 0 -1/J; 1/L 0]
C = [1 0 0]
D = [0 0]
motor_ss = ss(A,B,C,D)
%%% add system order
sys_order = order(motor_ss)
Example: DC Motor
>> motor_position
J =
3.2284e-006
b =
3.5077e-006
K =
0.0274
R =
4
L =
2.7500e-006
A =
1.0e+006 *
0 0.0000 0
0 -0.0000 0.0085
0 -0.0100 -1.4545
B =
1.0e+005 *
0 0
0 -3.0975
3.6364 0
C =
1 0 0
D =
0 0
a =
x1 x2 x3
x1 0 1 0
x2 0 -1.087 8487
x3 0 -9964 -1.455e+006
b =
u1 u2
x1 0 0
x2 0 -3.098e+005
x3 3.636e+005 0
c =
x1 x2 x3
y1 1 0 0
d =
u1 u2
y1 0 0
Continuous-time state-space model.
sys_order =
3
Control Oriented Model of
Batteries
Batteries:
State of charge: dx(t)/dt=k*u(t);
Open circuit voltage: voc(t)=f(x(t)) nonlinear
mapping of state of charge
Internal dynamics: distributed in nature
Modeling approach: lumped parameter model
vs distributed system
Control Oriented Model of
Batteries
Source: Lin, Stefanopoulou, Li, Anderson: Estimating Individual Cell States of Charge and Voltages
without Single Cell Voltage Measurement, Proceedings of 2013 American Control Conference
End of Week 1
Week 2
Syllabus
Policy Update
Quiz 1
HW1-4
Project
Adobe Acrobat
Document
Circuit Analysis
Mechanical System (Translational and Rotational)
Week 2 Contents
State Transition Matrix
] ) [( ) (
1 1

= = A sI L e t
At
Generalized Leibniz Integral Rule
Adjoint Matrix
Method: Inverse Laplace Transform
Hint: =t-
d = -d
Method:
A a A a I a A a A a a
A a A a A a A
I a A a A a A
I a A a A a A
n
n n
n
n
n
n
n
n
n
n
n
0
2
1 0 1
1
1 1
0
2
1 1
1
0 1
1
1
0 1
1
1
) (
0
=
=
=
= + + + +

L L
L
L
L
L L + + + + + =
n
n
A k A k A k I k A f
2
2 1 0
) (
any

=
+ + + + =
1
0
1
1
2
2 1 0
) (
n
k
k
k
n
n
A
A A A I A f

L
(

= =
1 0
2 1
?
100
A A
Example:
A I A A f let
1 0
100
) ( + = =
2 , 1 , 0 ) 2 )( 1 (
2 0
2 1
2 1
= = = =

100
2 1 0
100
2 2
100
1 1 0
100
1 1
2 ) (
1 ) (
= + = =
= + = =


f
f
1 2
2 2
100
1
100
0
=
=


=
(

+
(

= =
1 0
2 2 1
1 0
2 1
) 1 2 (
1 0
0 1
) 2 2 ( ) (
101
100 100 100
A A f
(


= =
0 2
1 3
? A e
At
Example:
2 , 1 , 0
2
1 3
2 1
= = =

2 ) 2 (
) 1 (
1 0 2 1 0
2
1 0 1 1 0


= + = =
= + = =

t
t
e f
e f
t t
t t
e e
e e


+ =
=
2
1
2
0
2

+ +

=
(


+ +
(

=



t t t t
t t t t
t t t t At
e e e e
e e e e
e e e e e
2 2 2
2
0 2
1 3
) (
1 0
0 1
2
2 2
2 2
2 2
Code Application of Cayley-Hamilton Theorem
% to find a^10
format long
a=[1 2; 0 3]
[t,j]=eig(a)
% adj(a)= no such command as
adj(a)=det(a)*inv(a)
adja=det(a)*inv(a)
colambda=[1 j(1,1); 1 j(2,2)]
flambda10=[j(1,1)^10; j(2,2)^10]
beta=inv(colambda)*flambda10
apower10=beta(1)*eye(2)+beta(2)*a
%
% a =
%
% 1 2
% 0 3
%
%
% t =
%
% 1.000000000000000
0.707106781186547
% 0 0.707106781186547
%
%
% j =
%
% 1 0
% 0 3
%
%
% adja =
%
% 3 -2
% 0 1
%
%
% colambda =
%
% 1 1
% 1 3
%
%
% flambda10 =
%
% 1
% 59049
%
%
% beta =
%
% -29523
% 29524
%
%
% apower10 =
%
% 1 59048
% 0 59049
Transpose of Cof
End of Week 2
Review of Week 2
Response of Linear Time-Invariant System
Calculation of e
At
Inverse Laplace Transform Method
Cayley- Hamilton Theorem Method
Transfer function from State Space Equations
Transition Matrix (Linear Time-Invariant and Linear Time-
Varying Systems)
Question about Quiz
Level of difficulty/Duration/Number of Questions
Not difficult
90 minutes so everyone has enough time to complete
Cover all main ideas discussed in the class.
Format: open book
What to take with you:
Pen/pencil
Calculator
Notes, books
Sheets of paper
Homework #1 is due as well
Recall the Buck Converter Circuits this time
we look at the Switch-On case
variables state as , Choose c L V i

+ =
+ + + =
+ =
) (
) (
* * ) (
) (
) (
* * ) ( *
) (
* ) (

) ( ) (
) (
0
0
t V
dt
t dV
C r t V
t V
dt
t dV
C r t i r
dt
t di
L t V
R
t V
dt
t dV
C t i
c
c
c
c
c
c
L
i
c
L
L L
analysis circuit on Based
( ) ( )
( ) ( )
( ) ( )

+
=
=
(

+
=
+

+
=
|
|

\
|
+

|
|

\
|
+
=

+
|
|

\
|
+ =
|
|

\
|
+ + =
+ =
) (
) (
) (
*
C * R
1
C * R
* * ) ( * ) (
) (
) (
* * ) ( * ) (
) (
*
) (
) (
*
C * R
1
C * R
) ( *
C * R
1
) ( *
C * R
) ( *
C * 1
1
) ( *
C * 1
1 ) (
) ( *
1

) (
* C * 1
) (
) (
* *
1

) (

) ( ) (
) (
0
t V
t V
t i
r r
R
C r t i r t V
t V
dt
t dV
C r t i r t V
dt
t di
L
t V
t i
r r
R
t V
r
t i
r
R
t V
R
r
R
t i
R
r dt
t dV
t V
R dt
t dV
R
r
t V
dt
t dV
C r
R dt
t dV
C
R
t V
dt
t dV
C t i
c
c
L
c c
c i
c
c
c i
L
c
L
c c
c
c
L
c
c
c
L
c
c
c
c c
c
c
c
c
c
L
L L
L L
( ) ( )
( ) ( )
( ) ( )
( ) ( )
( ) ( )
( ) ( )
(

+ +
=
(
(

+
(

(
(
(
(

+
+

+
+ +

=
(
(
(
(

+ +
=
(

+
=
=
+ =
(

+
+ +
+ =
) (
) (
*
R R
*
) (
) ( *
0
1
) (
) (
*
C * R
1
C * R
R *
R
R *
r * R r * r r * R
) (
) (
) (
) (
*
R R
*
) (
) (
*
C * R
1
C * R
* * ) ( *
) (
* * ) ( * ) (
) ( ) (
) (
) (
) (
*
R *
R
R *
r * R r * r r * R
) (
1 ) (
0
c c L L
0
0
c c L L
t V
t i
r
R
r
r R
t V
t V
L
t V
t i
r r
R
r L r L
dt
t dV
dt
t di
t V
t i
r
R
r
r R
t V
t i
r r
R
C R t i R
dt
t dV
C R t i R t V
R
t V
dt
t dV
C t i
t V
t i
r L r L
t V
L dt
t di
c
L
c c
c
i
c
L
c c
c c
c
L
c
L
c c
c
c
L
c c
L
c
L
c
L
c
L
c c
i
L

+ =
+ + + =
+ =
) (
) (
* * ) (
) (
) (
* * ) ( *
) (
* ) (

) ( ) (
) (
0
0
t V
dt
t dV
C r t V
t V
dt
t dV
C r t i r
dt
t di
L t V
R
t V
dt
t dV
C t i
c
c
c
c
c
c
L
i
c
L
L L
analysis circuit on Based
variables state as
dt
(t)
c
dV
and Choose c V
( ) ( ) ( )
[ ]
(

=
(
(

+
+
(

(
(

+
+ + +

+
+

=
(
(
(
(

= =
+ =
+
+
+ + +
+
+
=
+ +
|
|

\
|
|
|

\
|
+ + +
|
|

\
|
|
|

\
|
+ + =
+ + + =
|
|

\
|
+ + = + =
) (
) (
* * 1 ) (
) ( *
C * R *
0
) (
) (
*
C * R *
* ) * * * (
C * R *
1 0
) (
) (
) (
) ( ), ( ) ( Let
) (
) (
* * ) (
) ( *
) (
*
* ) * * * (
) ( * ) ( *
) (
) (
* * ) (
) (
* *
R
1
(t) X * C *
) (
) (
* *
R
1
(t) X * C *
) (
) (
* * ) ( *
) (
* ) (
) (
) (
* *
R
1
(t) X * C
) ( ) (
) (

2
1
0
2
1
2
1
2 1
0
2
2
2
2
2
0
t X
t X
C r t V
t V
r L
R
t X
t X
r L
C r r r R r R L
r L
r R
dt
t dX
dt
t dX
dt
t dV
t X t V t X
t V
dt
t dV
C r t V
t V
R
r R
dt
t dV
R
C r r r R r R L
dt
t V d
R
C r R L
t V
dt
t dV
C r t V
dt
t dV
C r r
t V
dt
t dV
C r
dt
d
L
t V
dt
t dV
C r t i r
dt
t di
L t V
t V
dt
t dV
C r
R
t V
dt
t dV
C t i
c
i
c c
c L c l
c
L
c
c
c
c
c
c
L c c L c l
c c
c
c
c c
c
c
c
c
c
c
c
c
L
i
c
c
c
c
L
L
L L
dt
t dV
t X t V t X
c
c
) (
) ( ), ( ) ( Let
2 1
= =
Notes:
Although two different state vectors can be selected, it is
obvious that the input and output relationship of the system
remains the same, hence the transfer function only reflects
the input/output relationship, not how the internal states are
represented. In other words, there may be many other
ways of selecting state variables.
For most engineering systems, they are far more
complicated than an LTI system. Still, LTI system
representation gives us a powerful approach to obtain
some of the basic properties of such systems.
Example: thermal system, battery dynamics
Notice that direct
feed-through term
does not change.
l=n-k n-(n-k)=k independent eigenvectors
l=n-1 n-(n-1)= 1 independent eigenvector
Use a 2x2 matrix as example:
( )
( )
[ ]
[ ] [ ]
[ ] [ ]
[ ] [ ] V1) of definition (by * *
* *
0
1
* *
0
1
*
0
1
T
*
0 *
1 ) ( and eigenvalue repeated h matrix wit 2x2 :
1 0 0 1 0
1 0 0 1 0
1 0 1 0
1
1 0
0 1
0
V V V V A V
V V V V A V A
V V V V A
T AT
AT T
V V
V V I A
V I A
A I rank A


+ =
+ =
(

=
(

=
(

=
=

=
=
=

Notice V1 is not unique


a=[1 -1 2; -2 5 4; -4 -5 8]
% a =
%
% 1 -1 2
% -2 5 4
% -4 -5 8
[t1,j]=eig(a)
% t1 =
%
% -0.8567 -0.2207 + 0.0714i -0.2207 - 0.0714i
% 0.1118 -0.1513 + 0.5861i -0.1513 - 0.5861i
% -0.5036 -0.7614 -0.7614
%
%
% j =
%
% 2.3063 0 0
% 0 5.8468 + 4.2243i 0
% 0 0 5.8468 - 4.2243i
t11=[t1(:,1) real(t1(:,2)) imag(t1(:,2))]
% t11 =
%
% -0.8567 -0.2207 0.0714
% 0.1118 -0.1513 0.5861
% -0.5036 -0.7614 0
inv(t11)*a*t11
% 2.3063 0.0000 0.0000
% 0.0000 5.8468 4.2243
% -0.0000 -4.2243 5.8468
The following is an implementation
For those extremely motivated to do some controls research later
1. Matrix Theory By Gantmacher, Vol. 1 & 2. TWO VOLUMES
2. C. T. Chen, Linear Systems Theory (intermediate)
3. W. M. Wonham, Linear Multivariable Control (difficult)
This image cannot currently be displayed.
Look at 3/(s+3)
Hint: Y(s)/U(s)=Y(s)/Z(s)*Z(s)/U(s)
This image cannot currently be displayed. This image cannot currently be displayed.
Why Controllable Canonical Form?
Why Controllable Canonical Form?
Why Controllable Canonical Form?
Example
The last notion is related to duality
between controllability and observability
which we shall explore later in Part 2.
1/22

Potrebbero piacerti anche