System Identification and Robust Control

Lecture 1: Introduction

Eugenio Schuster

schuster@lehigh.edu
Mechanical Engineering and Mechanics
Lehigh University

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 1 / 28
Book - System Identification

SYSTEM IDENTIFICATION
Theory for the User (Second Edition)
Lennart Ljung
Linköping University, Sweden

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 2 / 28
Content

1 Time-invariant Linear Systems [Chapter 2]

2 Nonparametric Time and Frequency Domain Methods [Chapter 6]

3 Simulation and Prediction [Chapter 3]

4 Models for Linear Time-invariant Systems [Chapter 4]

5 Parameter Estimation Methods [Chapter 7]

6 Convergence and Consistency [Chapter 8]

7 Computing the Estimate [Chapter 9]

8 Recursive Estimation Methods [Chapter 10]

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 3 / 28
Another Books

T. Söderström and P. Stoica, System Identification, Prentice Hall, 1989.

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 4 / 28
Book - Robust Control

MULTIVARIABLE FEEDBACK CONTROL

Analysis and design
Sigurd Skogestad
Norwegian University of Science and Technology
Ian Postlethwaite
University of Leicester

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 5 / 28
Content

1 Introduction [Chapter 1]

2 Classical feedback control [Chapter 2]

3 Performance limitations SISO [Chapter 5]

4 Uncertainty and robustness SISO [Chapter 7]

5 Elements of linear systems theory [Chapter 4]

6 Introduction to multivariable control [Chapter 3]

7 Performance limitations MIMO [Chapter 6]

8 Robust stability and performance MIMO [Chapter 8]

9 Controller design [Chapter 9]

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 6 / 28
Another Books

C. Chen, Linear System Theory and Design, Oxford University Press, 1999.

G. Dullerud and F. Paganini, A Course in Robust Control Theory, Springer,

2000.

M. Green and D. Limebeer, Linear Robust Control, Prentice Hall, 1995.

J. Helton and O. Merino, Classical Control Using H-infinity Methods, SIAM,

1998.

J. Maciejowski, Multivariable Feedback Design, Addison Wesley, 1989.

R. Sanchez-Pena and M. Sznaier, Robust Systems Theory and Applications,

John Wiley & Sons, 1998.

K. Zhou, Robust and Optimal Control, Prentice Hall, 1996.

K. Zhou, Essentials of Robust Control, Prentice Hall, 1998.

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 7 / 28
Web Page

http://www.lehigh.edu/~eus204/teaching/ME450_SIRC/ME450_SIRC.html

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 8 / 28
The Process of Control System Design

1 Study the “plant” to be controlled and learn about control objectives

2 Model the system and simplify the model (if necessary)
3 Analyze the resulting model; determine its properties
4 Decide which variables are to be controlled (controlled outputs)
5 Decide on measurement (sensors)/manipulated (actuators) variables
6 Select the control configuration
7 Decide on the type of controller to be used
8 Decide on performance specifications (based on control objectives)
9 Design a controller
10 Analyze resulting controlled system (redesign if necessary)
11 Simulate the resulting controlled system
12 Repeat from Step 2, if necessary.
13 Choose hardware and software and implement the controller
14 Test and validate control system and tune on-line if necessary

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 9 / 28
The Process of Control System Design
Time will be spent on input-output “controllability analysis” of the plant/process.

Always keep in mind

Power of control is limited.
Control quality depends controller AND on plant/process.

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 10 / 28
The Control Problem

y = Gu + Gd d (1)
y : output/controlled variable
u : input/manipulated variable
d : disturbance
r : reference/setpoint

Regulator problem : counteract d

Servo problem : let y follow r

Goal of control: make control error e = y − r “small”.

Note: To arrive at a good design for the controller K we need a priori

information about expected disturbances/references and knowledge of plant model
G and disturbance model Gd .

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 11 / 28
The Control Problem
Major difficulties: Model (G, Gd ) inaccurate ⇒ RealPlant: Gp = G + E ;

E = “uncertainty” or “perturbation” (unknown but bounded)

Nominal stability (NS) : system is stable with no model uncertainty

Nominal Performance (NP) : system satisfies performance specifications
with no model uncertainty
Robust stability (RS) : system stable for “all” perturbed plants
Robust performance (RP) : system satisfies performance specifications for
all perturbed plants

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 12 / 28
Transfer Functions
The course will be make extensive use of the transfer function representation, i.e.,
we will work on the frequency domain.

Given a state-space representation

ẋ = Ax + Bu,
y = Cx + Du,

the transfer-function representation is given by (assuming SISO system)

 
y(s) −1 A B
= G(s) = C(sI − A) B + D, G(s) = .
u(s) C D

For a MIMO system we write Y (s) = G(s)U (s).

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 13 / 28
Transfer Functions

βnz snz + · · · + β1 s + β0
G(s) = (2)
sn + an−1 sn−1 + · · · + a1 s + a0
For multivariable systems, G(s) is a matrix of transfer functions.
n = order of denominator (or pole polynomial) or order of the system
nz = order of numerator (or zero polynomial)
n − nz = pole excess or relative order.

Definition:
A system G(s) is strictly proper if G(s) → 0 as s → ∞.
A system G(s) is semi-proper or bi-proper if G(s) → D 6= 0 as s → ∞.
A system G(s) which is strictly proper or semi-proper is proper.
A system G(s) is improper if G(s) → ∞ as s → ∞.
Remark:
All practical systems have zero gain at sufficiently high frequency, and are
therefore strictly proper. It is often convenient, however, to model high-frequency
effects by a non-zero D-term, and hence semi-proper models are frequently used.
Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 14 / 28
Transfer Functions

Invaluable insights are obtained from simple frequency-dependent plots.

Important concepts for feedback such as bandwidth and peaks of closed-loop

transfer functions may be defined.

G(jω) gives the response to a sinusoidal input of frequency ω.

A series of interconnected systems corresponds in the frequency domain to

the multiplication of individual transfer functions, whereas in the time
domain, the evaluation of convolution operations is required.

Poles and zeros appear explicitly in factorized transfer functions.

Uncertainty is more easily handled in frequency domain.

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 15 / 28
Scaling
Proper scaling simplifies controller design and performance analysis.

SISO :
unscaled:
yb = Gb
bu + G
b d d;
b eb = yb − rb (3)
scaled:
d = d/
b dbmax , u=u
b/b
umax (4)
where:
dbmax — largest expected change in disturbance
u
bmax — largest allowed input change

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 16 / 28
Scaling
Scale yb, eb and rb by:
ebmax — largest allowed control error, or
rbmax — largest expected change in reference value

Usually:
y = yb/b
emax , r = rb/b
emax , e = eb/b
emax (5)

MIMO :
d = Dd−1 d,
b u = Du−1 u
b, y = De−1 yb (6)
e= De−1 eb, r= De−1 rb (7)
where De = ebmax , Du = u
bmax , Dd = dbmax and Dr = rbmax are diagonal scaling
matrices

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 17 / 28
Scaling
Substituting (6) and (7) into (3):

De y = GD
b uu + G
b d Dd d; De e = De y − De r

and introducing the scaled transfer functions

G = De−1 GD
b u, Gd = De−1 G
b d Dd (8)

Model in terms of scaled variables:

y = Gu + Gd d; e=y−r (9)

Often also:
rmax = Dr−1 rb
r̃ = rb/b (10)
so that: ∆
r = Rr̃ where R = De−1 Dr = rbmax /b
emax (11)

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 18 / 28
Scaling

d r̃

? ?
Gd R

r
u + y -
- G -?e -?e -e
+ +

Figure: Model in terms of scaled variables

Objective:
for |d(t)| ≤ 1 and |r̃(t)| ≤ 1,
manipulate u with |u(t)| ≤ 1
such that |e(t)| = |y(t) − r(t)| ≤ 1.

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 19 / 28
Linearization
Given the general nonlinear model

ẋ = f (x, u) (12)

where (x∗ , u∗ ) is an equilibrium, i.e., f (x∗ , u∗ ) = 0.

The linearization around this equilibrium is given by

 ∗  ∗
˙x̃ = ∂f x̃ +
∂f
ũ = Ax̃ + B ũ (13)
∂x ∂u

where x̃ = x − x∗ and ũ = u − u∗ . Therefore,

−1
x̃(s) = (sI − A) B ũ (14)

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 20 / 28
Notation
d
?
Gd
r-+b - K u- G - ?b + p -y
- +
ym 6
b+
+
6n
(a) One degree-of-freedom control configuration

d
?
Gd
r - u- G - ?b+ p - y
- K +
ym
b+
+
6n
(b) Two degrees-of-freedom control configuration

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 21 / 28
Notation

w - -z
- P

u v

K 

(c) General control configuration

Figure: Control configurations

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 22 / 28
Notation

Table: Nomenclature

K controller, in whatever configuration. Sometimes broken down into parts.

For example, in Figure 2(b), K = Kr Ky where Kr is a prefilter and
 

Ky is the feedback controller.

Conventional configurations (Fig 2(a), 2(b)):

G plant model
Gd disturbance model
r reference inputs (commands, setpoints)
d disturbances (process noise)
n measurement noise
y plant outputs. ( include the variables to be controlled (“primary” outputs
with reference values r) and possibly additional “secondary” measure-
ments to improve control)
ym measured y
u control signals (manipulated plant inputs)
Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 23 / 28
Notation

General configuration (Fig 2(c)):

P generalized plant model. Includes G and Gd and the interconnection
structure between the plant and the controller.
May also include weighting functions.
w exogenous inputs: commands, disturbances and noise
z exogenous outputs; “error” signals to be minimized, e.g. y − r
v controller inputs for the general configuration, e.g. commands, measured
plant outputs, measured disturbances, etc. For the special case of a
one degree-of-freedom controller with perfect measurements we have
v = r − y.
u control signals

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 24 / 28
How to Obtain Linear Models?
Linear models my be obtained from:
“First-principles” modeling. Linearization may be needed.
“Data-driven” modeling. Model identified from input-output data.
Combination of these two approaches.
Note that regardless of how the model is obtained, the model (G, Gd ) will always
be inaccurate:

RealPlant: Gp = G + E ;

E = “uncertainty” or “perturbation” (unknown but bounded)

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 25 / 28
System Identification

T : Sampling period; f = 1/T : Sampling rate/frequency (Hz)

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 26 / 28
System Identification
1. Identification of Discrete-time Model
Difference Equation:

y[k] + a1 y[k − 1] + · · · + an y[k − n] = bo u[k] + · · · + bm u[k − m]

Transfer Function (Z-Transform):

1 + a1 z + · · · + an z −n Y (z) = bo + · · · + bm z −m U (z)

Y (z) bo + · · · + bm z −m
G(z) = =
U (z) 1 + a1 z + · · · + an z −n

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 27 / 28
System Identification
2. Discrete-time → Continuous-time Model
Difference Equation → Differential Equation:

y (n) + b
a1 y (n−1) + · · · + b
an y = bbo u(m) + bb1 u(m−1) + · · · + bbm u

Transfer Function (Z-Transform) → Transfer Function (s-Transform):

 
sn + b
a1 sn−1 + · · · + b
an Y (s) = bbo sm + bb1 sm−1 + · · · + bbm U (s)

Y (s) bbo sm + bb1 sm−1 + · · · + bbm

G(s) = =
U (s) sn + b
a1 sn−1 + · · · + ban

READY FOR ROBUST CONTROL DESIGN!!!

Prof. Eugenio Schuster ME 450 - System Identification and Robust Control Spring 2020 28 / 28