Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
ECE/MAE 7360
Optimal and Robust Control
(Fall 2003 Offering)
ECE/MAE 7360 Optimal and Robust Control. Advanced methods of control system
analysis and design. Operator approaches to optimal control, including LQR/LQG/LTR,
mu-analysis, H-infinity loop shaping and gap metric etc. Prerequisite: ECE 6320 or
instructor approval. (3 cr) (alternate Fall).
Text: Kemin Zhou, with John Doyle, Essentials of Robust Control, Prentice-Hall, 1998.
Course Description: Robust control is concerned with the problem of designing control
systems when there is uncertainty about the model of the system to be controlled or when
there are (possibly uncertain) external disturbances influencing the behavior of the
system. Optimal control is concerned with the design of control systems to achieve a
prescribed performance (e.g., to find a controller for a given linear system that minimizes
a quadratic cost function). While optimal control theory was originally derived using the
techniques of calculus of variation, most robust control methodologies have been
developed from an operator-theoretic perspective. In this course we will mainly use an
operator approach to study the basic results in robust control that have been developed
over the last fifteen years. However, mathematical programming based techniques for
solving optimal control problems will also be briefly covered. This course provides a
unified treatment of multivariable control system design for systems subject to
uncertainty and performance requirements.
Course Topics:
From http://www.ece.usu.edu/academics/graduate_courses.html
***7360. Optimal and Robust Control.
Advanced methods of control system analysis and design. Operator approaches to optimal
control, including LQR, LQG, and L1 optimization techniques. Robust control theory, including
QFT, H-infinity, and interval polynomial approaches. Prerequisite: ECE/MAE 6320 or
instructor approval. Also taught as MAE 7360. (3 cr) (Sp)
Text: Kemin Zhou, with John Doyle, Essentials of Robust Control, Prentice-Hall, 1998.
References: Will be give by the Instructor via email/web/ftp.
Software: (1) MATLAB Control Systems Toolbox (2) MATLAB mu-Synthesis Toolbox (3)
RIOTS_95: MATLAB Toolbox for solving general optimal control problems.
Course Requirements:
Homework 40 points
Mid-term take home exam 10 points
Focused Individual Study Project/presentation 10 points
Design project 40 points
Notes:
Robust control is concerned with the problem of designing control systems when there is
uncertainty about the model of the system to be controlled or when there are (possibly uncertain)
external disturbances influencing the behavior of the system. Optimal control is concerned with
the design of control systems to achieve a prescribed performance (e.g., to find a controller for a
given linear system that minimizes a quadratic cost function). While optimal control theory was
originally derived using the techniques of calculus of variation, most robust control
methodologies have been developed from an operator-theoretic perspective. In this course we will
mainly use an operator approach to study the basic results in robust control that have been
developed over the last fifteen years. However, mathematical programming based techniques for
solving optimal control problems will also be briefly covered. This course provides a unified
treatment of multivariable control system design for systems subject to uncertainty and
performance requirements.
Course Topics:
Possible Topics for FISP (not limited to the following, students may propose their own topic
of interest subject to the Instructor’s approval)
1. l1 - and l∞ -optimal control (for rejection of unknown but bounded disturbances)
2. Structured perturbations, Kharitonov's Theorem
3. Quantitative feedback theory (QFT ).
4. Linear matrix inequalities (LMI ).
5. and many more …
3
Early years
Optimal control
• Bellman’s Dynamic Programming (1950’s)
• Pontryagin’s Maximum Principle (1950’s)
• Linear optimal control (late 1950’s and 1960’s)
– Kalman Filtering
– Linear-Quadratic (LQ) regulator problem
– Stochastic optimal control (LQG)
5
• British school
– Inverse Nyquist Array
– Characteristic Loci
• Singular values
– MIMO generalization of Bode gain plots
– MIMO generalization of Bode design
– Crude MIMO representations of uncertainty
• Multivariable loopshaping and LQG/LTR
– Attempt to reconcile modern and classical methods
– Popular, but hopelessly flawed
– Too crude a representation of uncertainty
Postmodern Control
• Mostly for fun. Sick of “modern control,” but wanted a name equally
pretentious and self-absorbed.
• Other possible names are inadequate:
– Robust ( too narrow, sounds too macho)
– Neoclassical (boring, sounds vaguely fascist )
– Cyberpunk ( too nihilistic )
• Analogy with postmodern movement in art, architecture, literature,
social criticism, philosophy of science, feminism, etc. ( talk about
pretentious ).
• linear subspaces
• eigenvalues and eigenvectors
• matrix inversion formulas
• invariant subspaces
• vector norms and matrix norms
• singular value decomposition
• generalized inverses
• semidefinite matrices
11
Linear Subspaces
• linear combination:
α1 x1 + . . . + αk xk , xi ∈ Fn, α ∈ F
span{x1, x2, . . . , xk } := {x = α1 x1 + . . . + αk xk : αi ∈ F}.
• x1, x2, . . . , xk ∈ Fn linearly dependent if there exists α1, . . . , αk ∈ F
not all zero such that α1 x2 + . . . + αk xk = 0; otherwise they are
linearly independent.
• {x1, x2, . . . , xk } ∈ S is a basis for S if x1, x2, . . . , xk are linearly
independent and S = span{x1, x2, . . . , xk }.
• {x1, x2, . . . , xk } in Fn are mutually orthogonal if x∗i xj = 0 for all
i 6= j and orthonormal if x∗i xj = δij .
• orthogonal complement of a subspace S ⊂ Fn :
S ⊥ := {y ∈ Fn : y∗x = 0 for all x ∈ S}.
• linear transformation
A: F
n
7−→ Fm.
• kernel or null space
KerA = N(A) := {x ∈ Fn : Ax = 0},
and the image or range of A is
ImA = R(A) := {y ∈ Fm : y = Ax, x ∈ Fn}.
Let ai, i = 1, 2, . . . , n denote the columns of a matrix A ∈ F
m×n
,
then
ImA = span{a1, a2, . . . , an}.
12
where yi ∈ Cn is given by
∗
y1
∗ −1
y2
1 x2 · · · xn
= x .
.
..
∗
yn
A11 A12 I 0 A11 0 I A−1
11 A12
• =
A21 A22 A21A−1
11 I 0 ∆ 0 I
∆ := A22 − A21A−1
11 A12
−1 ˆ 0
A11 A12 I A12 A22 ∆ I 0
• =
A21 A22 0 I 0 A22 A−1
22 A21 I
ˆ := A11 − A12A−1A21
∆ 22
−1
−1 −1 −1
A A A11 + A11 A12 ∆ A21A−1 −A−1 A12∆−1
• 11 12
= 11 11
A21 A22 −∆−1A21A−1 11 ∆−1
and
−1
A11 A12 ∆ˆ −1 −∆ˆ −1A12A−1
22
= −1 .
A21 A22 −A−1 A
22 21
ˆ
∆ −1
A−1
22 + A−1
A
22 21
ˆ
∆ −1
A A
12 22
−1
A11 0 A−1
11 0
=
A21 A22 −A−1 −1 −1
22 A21 A11 A22
−1
−1
A11 A12 A11 −A−1 −1
11 A12 A22
= .
0 A22 0 A−1
22
(A11 − A12A−1
22 A21 )
−1
= A−1 −1 −1 −1 −1
11 + A11 A12 (A22 − A21A11 A12 ) A21 A11 .
16
Invariant Subspaces
• a subspace S ⊂ C
n
is an A-invariant subspace if Ax ∈ S for every
x ∈ S.
For example, {0}, Cn , and KerA are all A-invariant subspaces.
Let λ and x be an eigenvalue and a corresponding eigenvector of
A ∈ Cn×n . Then S := span{x} is an A-invariant subspace since
Ax = λx ∈ S.
In general, let λ1, . . . , λk (not necessarily distinct) and xi be a set of
eigenvalues and a set of corresponding eigenvectors and the generalized
eigenvectors. Then S = span{x1, . . . , xk } is an A-invariant subspace
provided that all the lower rank generalized eigenvectors are included.
• An A-invariant subspace S ⊂ Cn is called a stable invariant subspace
if all the eigenvalues of A constrained to S have negative real parts.
Stable invariant subspaces are used to compute the stabilizing solu-
tions of the algebraic Riccati equations
• Example
λ1 1
λ1
A x1 x2 x3 x4 = x1 x2 x3 x4
λ3
λ4
with Reλ1 < 0, λ3 < 0, and λ4 > 0. Then it is easy to verify that
S1 = span{x1} S12 = span{x1, x2} S123 = span{x1, x2, x3}
S3 = span{x3} S13 = span{x1, x3} S124 = span{x1, x2, x4}
S4 = span{x4} S14 = span{x1, x4} S34 = span{x3, x4}
are all A-invariant subspaces. Moreover, S1, S3, S12, S13 , and S123 are
stable A-invariant subspaces.
17
n
X
kAk∞ = max |aij | (row sum) .
1≤i≤m j=1
A∗ Avi = σi2vi
AA∗ ui = σi2ui .
Generalized Inverses
Semidefinite Matrices
• F. L. Lewis, Applied Optimal Control and Estimation, Prentice Hall, Englewood Cliffs,
New Jersey, 1992.
• A. Saberi, B. M. Chen and P. Sannuti, Loop Transfer Recovery: Analysis and Design,
Springer, London, 1993.
• A. Saberi, P. Sannuti and B. M. Chen, H2 Optimal Control, Prentice Hall, London, 1995.
3
Prepared by Ben M. Chen
Revision: Basic Concepts
6
Prepared by Ben M. Chen
What is a control system?
Objective: To make the system OUTPUT and the desired REFERENCE as close
as possible, i.e., to make the ERROR as small as possible.
Desired Government
Economic System
Performance Policies
8
Prepared by Ben M. Chen
A Live Demonstration on Control of a Coupled-Tank System through Internet Based
Virtual Laboratory Developed by NUS
The objective is to control the flow levels of two coupled tanks. It is a reduced-scale
model of some commonly used chemical plants.
9
Prepared by Ben M. Chen
Modelling of Some Physical Systems
By the well-known Newton’s Law of motion: f = m a, where f is the total force applied to an
object with a mass m and a is the acceleration, we have
b u
u − bx& = m&x& ⇔ &x& + x& =
m m
This a 2nd order Ordinary Differential Equation with respect to displacement x. It can be
written as a 1st order ODE with respect to speed v = x& :
b u ← model of the cruise control system, u is input force, v is output.
v& + v= 10
m m
Prepared by Ben M. Chen
A cruise-control system:
90 km/h u b u speed v
? v& + v =
+ m m
–
11
Prepared by Ben M. Chen
Basic electrical systems:
i i (t) i (t)
di
v R v=iR v (t) C i=C
dv v (t) L v=L
dt
dt
The sum of voltage drops around any The sum of currents entering/leaving a
close loop in a circuit is 0. note/closed surface is 0.
v2 i1 i1
i5 i5
v1 v3 i4 i4
i2 i2
i3 i3
v1 +v2 +v3 +v4 +v5 = 0
v5 v4 i1 + i2 + i3 + i4 + i5 = 0
12
Prepared by Ben M. Chen
Modelling of a simple electrical system:
To find out relationship between the input (vi) and the output (vo) for the circuit:
dvo
vR = Ri = RC
dt
i dvo
i=C
dt
R
vi C vo
By KVL, we have v o + v R − vi = 0
dvo
vo + v R − vi = vo + RC − vi = 0
dt
230 Volts vi vo
? RC v&o + vo = vi
+
–
14
Prepared by Ben M. Chen
Ordinary Differential Equations
Many real life problems can be modelled as an ODE of the following form:
&y&(t ) + a1 y& (t ) + a0 y (t ) = u (t )
This is called a 2nd order ODE as the highest order derivative in the equation is 2. The ODE
is said to be homogeneous if u(t) = 0. In fact, many systems can be modelled or
approximated as a 1st order ODE, i.e.,
y& (t ) + a0 y (t ) = u(t )
An ODE is also called the time-domain model of the system, because it can be seen the above
equations that y(t) and u(t) are functions of time t. The key issue associated with ODE is: how
to find its solution? That is: how to find an explicit expression for y(t) from the given equation?
15
Prepared by Ben M. Chen
State Space Representation
Recall that many real life problems can be modelled as an ODE of the following form:
&y&(t ) + a1 y& (t ) + a0 y (t ) = u (t )
x1 = y x&1 = y& = x2
x2 = y& x&2 = &y& = − a1 y& − a0 y + u = −a1 x2 − a0 x1 + u
x&1 0 1 x1 0 x1
= + u, y = x1 = [1 0]
x&2 − a0 − a1 x2 1 x2
This is called the state space representation of the ODE or the dynamic systems.
16
Prepared by Ben M. Chen
Laplace Transform and Inverse Laplace Transform
0.8
1.5
0.6
1
0.4
0.5
0.2
Magnitude
Magnitude
0 0
-0.2
-0.5
-0.4
-1
-0.6
-1.5
-0.8
-1 -2
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
TIME (Second) TIME (Second)
∞ ∞ ∞
1 1 1 1 1 1
F ( s) = ∫ f (t ) e dt = ∫ e dt = − e − st = − e −∞ − − e0 = − ⋅ 0 − − ⋅ 1 = , Re( s ) > 0
− st − st
0 0 s 0 s s s s s
∞ ∞ ∞ ∞
1 − ( s + a )t 1
F (s) = ∫ f (t )e − st dt = ∫ e −at e − st dt = ∫ e −( s + a )t dt = − e = , Re( s ) > −a
0 0 0 s + a 0 s + a
18
Prepared by Ben M. Chen
Inverse Laplace Transform
Given a frequency-domain function F(s), the inverse Laplace transform is to convert it back
to its original time-domain function f (t).
Here are some very useful Laplace and inverse Laplace transform pairs:
f (t ) ⇔ F ( s) f (t ) ⇔ F ( s)
1 a
1 ⇔ sin at ⇔
s s2 + a 2
1 s
t ⇔ cos at ⇔
s2 s2 + a 2
1 b
e−at ⇔ e−at sin bt ⇔
s+a (s + a )2 + b2
1 s+a
te−at ⇔ e −at cos bt ⇔
(s + a )2 (s + a )2 + b2
19
Prepared by Ben M. Chen
Some useful properties of Laplace transform:
1. Superposition:
df (t )
L { }
= L f& (t ) = sL{ f (t )} = sF ( s )
dt
d 2 f (t )
L 2
{ }
= L &f&(t ) = s L{ f (t )} = s F ( s )
2 2
dt
3. Integration:
t 1
L ∫ f (ζ )dζ = L{ f (t )} = F ( s )
1
0 s s
20
Prepared by Ben M. Chen
Re-express ODE Models using Laplace Transform (Transfer Function)
Recall that the mechanical system in the cruise-control problem with m = 1 can be
represented by an ODE:
v& + bv = u
⇒ (s + b) V ( s) = U ( s) ⇒
V ( s)
=
1
= G (s )
U ( s) s+b
22
Prepared by Ben M. Chen
In general, a feedback control system can be represented by the following block diagram:
Given a system represented by G(s) and a reference R(s), the objective of control system
design is to find a control law (or controller) K(s) such that the resulting output Y(s) is as
close to reference R(s) as possible, or the error E(s) = R(s) –Y(s) is as small as possible.
However, many other factors of life have to be carefully considered when dealing with real-
life problems. These factors include:
uncertainties
disturbances noises
R (s) E (s)
K (s) G (s)
+ – U (s) Y (s)
nonlinearities
23
Prepared by Ben M. Chen
Control Techniques – A Brief View:
There are tons of research published in the literature on how to design control laws for various
purposes. These can be roughly classified as the following:
♦ Optimal control: Linear quadratic regulator control, Kalman filter, H2 control, developed in
1960s to achieve certain optimal performance and boomed by NASA Apollo Project.
♦ Robust control: H∞ control, developed in 1980s & 90s to handle systems with uncertainties
and disturbances and with high performances. Example: military systems.
♦ Nonlinear control: Currently hot research topics, developed to handle nonlinear systems
with high performances. Examples: military systems such as aircraft, missiles.
♦ Intelligent control: Knowledge-based control, adaptive control, neural and fuzzy control, etc.,
researched heavily in 1990s, developed to handle systems with unknown models.
Examples: economic systems, social systems, human systems. 24
Prepared by Ben M. Chen
Classical Control
Recall that the objective of control system design is trying to match the output Y(s) to the
reference R(s). Thus, it is important to find the relationship between them. Recall that
Y ( s)
G (s) = ⇒ Y ( s ) = G ( s )U ( s )
U ( s)
Similarly, we have U ( s ) = K ( s ) E ( s ), and E ( s ) = R ( s ) − Y ( s ) . Thus,
Y ( s ) = G ( s )U ( s ) = G ( s ) K ( s ) E ( s ) = G ( s ) K ( s ) [R ( s ) − Y ( s ) ]
Y ( s ) = G ( s ) K ( s ) R ( s ) − G ( s ) K ( s )Y ( s ) ⇒ [1 + G ( s ) K ( s ) ] Y ( s ) = G ( s ) K ( s ) R ( s )
Y (s) G (s) K (s)
⇒ H ( s) = = Closed-loop transfer function from R to Y.
R(s) 1 + G (s) K (s) 25
Prepared by Ben M. Chen
Thus, the block diagram of the control system can be simplified as,
The whole control problem becomes how to choose an appropriate K(s) such that the
resulting H(s) would yield desired properties between R and Y.
b
We’ll focus on control system design of some first order systems G ( s ) = with a
s+a
ki k p s + ki
proportional-integral (PI) controller, K ( s ) = k p + = . This implies
s s
G (s) K (s) bk p s + bk i
H (s) = = 2
1 + G ( s ) K ( s ) s + ( a + bk p ) s + bk i
The closed-loop system H(s) is a second order system as its denominator is a polynomial s
of degree 2.
26
Prepared by Ben M. Chen
Stability of Control Systems e − at ⇔
1
s+a
Example 1: Consider a closed-loop system with,
0.5
R (s) = 1 Y (s) 0.5 e −t ⇔
1 s +1
H (s) =
s2 − 1 &
We have 0.5 et ⇔
0.5
s −1
1 1 0 .5 0 .5
Y (s) = H ( s)R (s) = = = −
s2 − 1 ( s + 1)( s − 1) s −1 s +1
12000
2000
s = 1.
0
0 2 4 6 8 10
Time (seconds) 27
Prepared by Ben M. Chen
Example 2: Consider a closed-loop system with,
R (s) = 1 1 Y (s)
H (s) = 2
s + 3s + 2
We have
1 1 1 1
Y (s) = H (s)R(s) = = = −
s + 3s + 2
2
( s + 1)( s + 2 ) s +1 s+2
0.25
0
0 2 4 6 8 10
Time (seconds)
28
Prepared by Ben M. Chen
We consider a general 2nd order system,
R (s) = 0 Y (s)
ω n2
H (s) = 2
s + 2ζω n s + ω n2
The system is stable if the denominator of the system, i.e., s 2 + 2ζω n s + ω n2 = 0 , has no
positive roots. It is unstable if it has positive roots. In particular,
Marginally Stable
Stable
Unstable
29
Prepared by Ben M. Chen
Stability in the State Space Representation
x& = A x + B u
y =C x + Du
Then,
1. It is stable if and only if all the eigenvalues of A are in the open left-half plane.
2. It is marginally stable if and only if A has eigenvalues are in the closed left-half
plane with some (simple) on the imaginary axis.
3. It is unstable if and only if A has at least one eigenvalue in the right-half plane.
L.H.P. R.H.P.
30
Prepared by Ben M. Chen
Lyapunov Stability
Consider a general dynamic system, x& = f (x ) . If there exists a so-called Lyapunov function
V(x), which satisfies the following conditions:
V ( x ) → ∞, as x → ∞
then we can say that the system is globally asymptotically stable at x = 0. In this case, the
stability is independent of the initial condition x(0).
31
Prepared by Ben M. Chen
Lyapunov Stability for Linear Systems
Consider a linear system, x& = A x . The system is asymptotically stable (i.e., the eigenvalues
of matrix A are all in the open RHP) if for any given appropriate dimensional real positive
definite matrix Q = QT > 0, there exists a real positive definite solution P = PT > 0 for the
following Lyapunov equation:
AT P + PA = −Q
Proof. Define a Lyapunov function V ( x ) = x T P x . Obviously, the first and second conditions
on the previous page are satisfied. Now consider
≥ 0 and A, Q 2 being
1
Note that the condition, Q = QT > 0, can be replaced by Q = QT
detectable. 32
Prepared by Ben M. Chen
Behavior of Second Order Systems with a Step Inputs
Again, consider the following block diagram with a standard 2nd order system,
33
Prepared by Ben M. Chen
Control System Design with Time-domain Specifications
overshoot M p
rise time
1 .8 tr
tr ≅ t
ωn ts ts ≅
4 .6
1% settling time ζω n
34
Prepared by Ben M. Chen
PID Design Technique:
b k k s + ki
with G ( s ) = and K ( s ) = k p + i = p results a closed-loop system:
s+a s s
Y ( s) G ( s)K ( s) bk p s + bk i
H (s) = = = 2
R ( s ) 1 + G ( s ) K ( s ) s + ( a + bk p ) s + bk i
To be safe, we choose ζ = 0 . 6
4 .6 4 .6 4 .6
ts = ⇒ ωn = = = 0 . 767
ζω n ζ ts 0 . 6 × 10
36
Prepared by Ben M. Chen
The transfer function of the cruise-control system,
1 1
Y (s) m = 3000
G( s) = = ⇒ a=b= 1 = 0.000333
U ( s) s + b s+ 1 3000
m 3000
Again, using the formulae derived,
Reference
90 km/h Speed
1765
2760 +
+ s
–
37
Prepared by Ben M. Chen
Simulation Result:
120
The resulting
100 overshoot is
less than 25%
80
60 time is about 10
seconds.
40
Thus, our
20
design goal is
achieved.
0
0 2 4 6 8 10 12 14 16 18 20
Tim e in S e c o n d s
38
Prepared by Ben M. Chen
Bode Plots
r e y r e y
K (s) G(s) K(s) G(s)
+ – + –
Bode Plots are the the magnitude and phase responses of the open-loop transfer function,
i.e., K(s) G(s), with s being replaced by jω. For example, for the ball and beam system we
considered earlier, we have
3 .7 + 2 .3s 3 . 7 + j 2 . 3ω
K ( s ) G ( s ) s = jω = (0 . 37 + 0 . 23 s )
10
= =
s2 s = jω s2 s = jω −ω2
3 . 7 2 + ( 2 . 3ω ) 2 2 . 3ω
K ( j ω )G ( j ω ) = , ∠ K ( j ω ) G ( j ω ) = tan −1 − 180
o
ω 2
3 .7
39
Prepared by Ben M. Chen
Bode magnitude and phase plots of the ball and beam system:
60
40
Magnitude (dB)
20
-20
-1 0 1
10 10 10
Frequency (rad/sec)
-80
-100
Phase (degrees)
-120
-140
-160
-180
-1 0 1
10 10 10
Frequency (rad/sec)
40
Prepared by Ben M. Chen
Gain and phase margins
20
gain
0 margin
Magnitude (dB)
-20
-40
-60
-1 0 1
10 10 10
gain
Frequency (rad/sec)
crossover
frequency
0 phase
-50
crossover
frequency
Phase (degrees)
-100
-150
-200
phase
margin
-250
-1 0 1
10 10 10
Frequency (rad/sec)
41
Prepared by Ben M. Chen
Nyquist Plot
Instead of separating into magnitude and phase diagrams as in Bode plots, Nyquist plot
maps the open-loop transfer function K(s) G(s) directly onto a complex plane, e.g.,
1.5
0.5
Imag Axis
-0.5
-1
-1.5
-0.5 0 0.5 1 1.5
Real Axis
42
Prepared by Ben M. Chen
Gain and phase margins
The gain margin and phase margin can also be found from the Nyquist plot by zooming in
the region in the neighbourhood of the origin.
1
GM = , where ω p is such that ∠ K ( jω p )G ( jω p ) = 180o
K ( jω p )G ( jω p )
60
40
Magnitude (dB)
20
-20
-1 0 1
10 10 10
Frequency (rad/sec)
-80
-100
Phase (degrees)
-120
-140
-160
-180
-1 0 1
10 10 10
Frequency (rad/sec)
44
Prepared by Ben M. Chen