METR4202/7202 – Advanced Control

Lecture 2

Material from Various Sources,

Mainly Nise Chapters 5 and 6
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 1

Introduction to State Space Modelling

and Controller Design

G. Hovland 2004-2006

Why State-Space Methods

Opens up control possibilities for
Nonlinear systems
Multiple-input, multiple-output systems
Elegant Observer Design (such as the Kalman Filter)
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 2
 State Space Modelling - Aims

Why use “state space”

What is “state space”
How to form state space models
How to convert a transfer function to state space and
vice versa
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 3

How to solve the state equations using Laplace and the

implications for stability analysis

State Space Modelling - Aims (cont’d)

The analytical, continuous time solution for the time

response.
The computer simulation solution for the time response.
How to represent a state space system as a signal flow
graph.
How the signal flow graph can be used to convert
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 4

between alternate representations in state space.

 Classic vs. Modern Control

Control System Design

state-variable methods  modern  state space
root locus & frequency response  classical  transfer function
Design objective
same in both methods e.g. tracking, regulation
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 5

in situations where state-variables can be measured, state-

space design methods are more powerful

Advantages of state-space design become obvious in the relative

simplicity with which multi-input multi-output (MIMO) systems can
be treated.
Also state-variable representation provides information on internal
representation not available from only input-output data.

System & State Variables

System variable: any variable that responds to an input

or initial conditions in a system.
State variable: the smallest set of linearly independent
system variables such that the values of the members of
the set at time t0 along with known forcing functions
completely determine the value of all system variables
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 6

for all t ≥ t0.

Inverted pendulum states: two angles q1, q2 and two
velocities
d/dt(q1), d/dt(q2)
 State Space

Is the set or space containing the states

Idea came from
state-variable description of differential equations
a rewriting of a high order differential equation into
first order form
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 7

Control system design using the state-variable

description is called state-space design

Example 1

Given the second-order system:

y + a1 y + a 2 y = bu
It can be rewritten as (phase-variable form):
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 8

x1 = y
x1 = x2
x2 = x1 = y
x2 = − a 2 x1 − a1 x2 + bu
x2 = y
X = AX + Bu
 Example 2: State Space Representation

State variables: loop current i(t) and capacitor charge

q(t)

R L i(t)

+ C
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 9

Vc(t)
di 1 V
L + Ri + idt = V −
dt C

dq
i=
dt

Example 2: State Space Representation

R L i(t)

+ C
Vc(t)
V
di 1
L + Ri + idt = V −
dt C
di − 1 −R 1
= q+ i+ V q
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 10

dt LC L L x=
i x = Ax + Bu
dq u V y = Cx + Du
i= y VL
dt
di − 1
VL = L = q − Ri + V
dt C
! "! " # $
! %&%' ()
 Example 2: State Space Representation

di − 1 −R 1
= q+ i+ V q
dt LC L L x=
i x = Ax + Bu
dq u V y = Cx + Du
i= y VL
dt
di − 1
VL = L = q − Ri + V
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 11

dt C
x1 0 1 x1 0
R L
= −1 −R
+ 1 u
i(t) x2 LC L x2 L

x1
y =[ − R] + [1]u
+ C
Vc(t) −1
V C
−
x2

Matlab Simulation of Example 2

!" #
$ $% & '
"$ ( $
$)
$* '+ ,
-
& #
$! . + / "
!. + %
0 #
$ +%
#
. + !. % 1
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 12

2 . 3
( 3 , 4
5 $$$ 1- $$$
&/ ! 5"6 0/-"/ 7 1
V oltage ac ros s induc tor

!
56 " !5"6 0.8

56 " 5"6 0.6

0.4
V oltage (V )

/ 6 2/- 1 - 4
0.2
1
!" 0

, -0.2

8
9 , 8
" -0.4
0 2 4 6 8 10 12 14

4 8( "8
"
Tim e (s )

4 89 , 9"8
"
&!0! !2"
"
 Simulink Implementation of State-Space Models
Matrix Gain
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 13

*+ *, Alternatives:

*,

SISO or MIMO SISO

Some reasons for using state-variable form

to study more general models

state-variable can easily include non-linear as well as time-varying
systems.
Time-varying systems

y + a1 (t ) y + a2 (t ) y = b(t )u
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 14

x1 = y , x1 = y = x2
x2 = y , x2 = y = − a 2 (t ) x1 − a1 (t ) x2 + b (t )u
where transfer function description is not possible state variable
description is straightforward.
 Some reasons for using state-variable form
Introduce ideas of geometry into differential equations.
plot of position vs velocity is a phase plane, which shows motion
of system in two-dimension.
State-space generalises this idea to higher dimensions.
Visualise solutions of differential equations as trajectory in state-
space of appropriate dimension.
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 15

to connect internal and external descriptions

state-variables show the internal dynamics of the system.
Through the matrix representation, a simple relationship can be
established to the external input-output description.
Transfer-function description focuses on input-output
representation only

A power system example

% " .
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 16

- !
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 18 METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 17

A power system example: Phase Portrait 2 – Stable

A power system example: Phase portrait I - Oscillations
 Visualisation - Computer Graphics

State variables can be used to visualise a mechanical

system.

Prac 1: Inverted Pendulum

METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 19

Swing-Up Controller: PD Controller

Balancing Controller: State-Space

Switching Logic: Stateflow

Controller Platform: xPC

Forming the State-Space Representation

Select State Variables:

at least the minimum number of state variables =
order of the differential equation describing the
system.
all state variables must be linearly independent of
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 20

each other ⇔ one variable can not be linear

combination of other variable(s).
 Forming the State-Space Representation

State equations:
! "! " # $
! %&%' ()

x1 = a11 x1 + a21 x2 + ... + an1 xn + b1u

x2 = a12 x1 + a22 x2 + ... + an 2 xn + b2 u
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 21

.
⇔ x = Ax + Bu

xn = a1n x1 + a2 n x2 + ... + ann xn + bn u

Output equation(s): y = Cx + Du

Converting a Transfer Function into State Space

Y ( s) b0
Transfer function: =
U ( s ) ( s n + an −1s n −1 + a1s + a0 )
Assume first only
Cross-multiply: poles, no zeros

( s n + a n −1 s n − 1 + + a 1 s + a 0 )Y ( s ) = b0U ( s )
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 22

d
Remember: s ↔
dt

take the inverse Laplace transform (assuming zero

initial conditions):
dny d n −1 y dy
+ a n −1 n −1 + + a1 + a0 y = b0 u
dt n dt dt
 Converting a Transfer Function into State Space
(phase-variable)

dny d n −1 y dy
n
+ an −1 n −1 + + a1 + a0 y = b0 u
dt dt dt

/ x1=y 0
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 23

x1 = x2
x2 = x3

xn −1 = xn
xn = − a0 x1 − a1 x2 − an −1 xn + b0u

Converting a Transfer Function into State Space

(phase-variable)
Y (s) b0
= n
U ( s ) ( s + a n −1 s n −1 + a1 s + a0 s )

% !
x1 0 1 0 0 x1 0
x1=y x2 0 0 1 0 x2 0
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 24

x1 = x2 = + u
x2 = x3 xn−1 0 0 0 1 xn−1 0
xn − a0 − a1 − an−1 xn 1

xn −1 = xn
xn = − a0 x1 − a1 x2 − an −1 xn + b0u
 Convert Transfer Function into State Space: Y/U =
b(s)/a(s)

Next, we introduce the zeros

X 1( s) 1
= n
U ( s ) s + an −1s n −1 + + a1s + a0 s
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 25

Y ( s)
= bn −1s n −1 + + b1s + b0
X 1( s)

y (t ) = bn −1 xn + + b1 x2 + b0 x1
1

Convert Transfer Function into State Space: Y/U =

b(s)/a(s)

y (t ) = bn −1 xn + + b1 x2 + b0 x1
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 26

x1
x2
y = [b0 b1 bn−1 ]

xn
 Convert Transfer Function into State Space - Summary

Y ( s) bn −1s n −1 + + b1 ( s ) + b0
=
U ( s ) s n + an −1s n −1 + + a1s + a0 s
x1 0 1 0 0 x1 0
x2 0 0 1 0 x2 0
= + u
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 27

xn −1 0 0 0 1 xn −1 0
xn − a0 − a1 − an −1 xn 1
x1
A matrix B vector
x2
y = [b0 b1 bn−1 ]
D=0
C vector xn

Transfer Function - Block diagram – State Space

A dynamic system described by differential equations

can be represented by transfer function;
And can be converted into state space via block diagram
(with fundamental building block as integrator unit)
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 28

Transfer State-Space
Function Model

Input-output behaviour is identical, however,

the underlying model structure is lost with transfer functions
 Convert State Space into Transfer Function

Using Laplace transform:

x = Ax + Bu 1

y = Cx + Du
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 29

Y ( s)
T ( s) = = C( sI − A ) −1 B + D
U ( s)

In Matlab use tf2ss and ss2tf functions

Next Topics

Laplace solution to state equation & stability

Stability in State Space
Signal flow graphs
Alternative representations in state space
Phase-variable form
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 30

Cascade form
In Lecture 3 and 4 we will
Parallel form design controllers and
Controller canonical form observers based on these
two forms
Observer canonical form
 Laplace Solution to State Equations and Stability

x = Ax + Bu
y = Cx + Du
/ 2 !
sX ( s ) − X (0) = AX ( s ) + BU ( s )
Y ( s ) = CX ( s ) + DU ( x)
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 31

3 0 4! *#( "5#(
X ( s ) = ( sI − A) −1[ X (0) + BU ( s )]
adj ( sI − A )
= [ X (0) + BU ( s )]
det( sI − A )
Y ( s ) = CX ( s ) + DU ( s )

Laplace Solution to State Equations and Stability (sI-A)-1=adj(sI-A)/det(sI-A)

46 " #
*#7( 7(

Y (s) adj( sI − A)
T ( s) = =C B+D
U (s) det( sI − A)
C adj( sI − A )B + D det( sI − A )
=
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 32

det( sI − A)
Poles of a state-space system are formed by:
Roots of the denominator, i.e. det(sI-A)
The eigenvalues of the system matrix A, i.e. det(sI-A)=0.

In Matlab: [num,den]=ss2tf(A,B,C,D,1)
roots(den)
eig(A)
 State Space Representation to Transfer function e.g. 3.6

0 1 0 10
X= 0 0 1 X+ 0 u
−1 − 2 − 3 0
y = [1 0 0]X
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 33

m11 m12 m13 k11 k12 k13

adj ( m21 m22 m23 ) = k 21 k 22 k 23
adj ( sI − A) m31 m32 m33 k31 k32 k33

s −1 0 kij = ( −1)i + j det( M ji ) )4)

k11 = (−1)1+1 ⋅ (m22 m33 − m23 m32 )
= adj 0 s −1
= s ⋅ ( s + 3) − ( −1 ⋅ 2) = s 2 + 3s + 2
1 2 s+3 k12 = ( −1)1+ 2 ⋅ [ −1 ⋅ ( s + 3) − 0 ⋅ 2] = s + 3

State Space Representation to Transfer function e.g. 3.6

m11 m12 m13 k11 k12 k13
adj ( m21 m22 m23 ) = k 21 k 22 k 23
0 1 0 10
m31 m32 m33 k31 k32 k33
X= 0 0 1 X+ 0 u
−1 − 2 − 3 0 kij = ( −1)i + j det( M ji ) )4)
y = [1 0 0]X k11 = (−1)1+1 ⋅ (m22 m33 − m23 m32 )
= s ⋅ ( s + 3) − ( −1⋅ 2) = s 2 + 3s + 2

3 k12 = ( −1)1+ 2 ⋅ [ −1 ⋅ ( s + 3) − 0 ⋅ 2] = s + 3
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 34

s 0 0 0 1 0 s −1 0 10
8( ( sI − A) = 0 s 0 − 0
# 0 1 = 0 s −1 B= 0
0 0 s −1 − 2 − 3 1 2 s+3 #
:( 0
s 2 + 3s + 2 s+3 1 C = [1 0 0]
−1 s ( s + 3) s
#
;(
−s − (2 s + 1) s 2 10( s 2 + 3s + 2)
#
9( ( sI − A) −1 =
adj ( sI − A)
= T (s) = 3
det( sI − A) s 3 + 3s 2 + 2 s + 1 s + 3s 2 + 2 s + 1
Y (s)
T (s) = = C( sI − A) −1 B + D
U (s)
 Laplace transform solution of state equations – exercise 4.9

Solve for y(t) and find eigenvalues and system poles.

0 2 0 u = e −t
X= X+ u
−3 −5 1 2
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 35

y = [1 3]X X ( 0) =
1

Laplace transform solution of state equations – exercise 4.9

Solve for y(t) and find eigenvalues and system poles.

0 2 0 u = e−t
X= X+ u
−3 −5 1 2
y = [1 3]X X ( 0) =
1
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 36
 Matlab simulation

The system is stable:

t=[0:0.01:10];
y=-0.5.*exp(-t)-12.*exp(-2*t)+17.5.*exp(-3*t);
plot(t,y)
grid
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 37

Stability in State Space

Ax = λx or (λI − A) x = 0
λ : eigenvalue
x: eigenvector corresponding to eigenvalue λ
MATLAB: [x, λ] = eig(A) diagonal matrix λ of
eigenvalues and a full matrix x whose columns are
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 38

the corresponding eigenvectors

Stability Conditions for x’=Ax+Bu, y=Cx+Du
If all eigenvalues of A are located in the left hand side
of the s-plane, the system is stable.
Otherwise, the system is unstable
 Routh Table – revision* characteristic equations of closed-
loop TF: a4s4+a3s3+a2s2+a1s+a0=0

1 6
<) ! !

4 !
$ !4
4 !
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 39

)
< 4 4 8
⇔ $)

Table 6.2 Completed Routh table

Stability in state space – eigenvalue vs Routh Table e.g.6.11

=" 0 3 1 10
X= 2 8 1 X+ 0 u
! 4 ! "> )
− 10 − 5 − 2 0
y = [1 0 0]X

s 0 0 0 3 1 s −3 −1
( sI − A) = 0 s 0 − 2 8 1 = − 2 s − 8 −1
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 40

0 0 s − 10 − 5 − 2 10 5 s+2
det( sI − A) = s 3 − 6 s 2 − 7 s − 52
2 $ @4 0
s3 1 -7 4#( 0 ! A& A 7
s2 -6 -3 -52 -26 4 0 −0.8821 ±
?
j2.4330, 7.7642 4 0
s1 -47/3 -1 0 4 ! $)
s0 -26

8 4 4 8 4 ! $)
 Signal Flow Graphs of State Equations

An alternative representation to the block diagram form used for transfer

function representations.
Nodes : signals or variables
Branches: transfer functions
All branches are summed at a node
Output is formed from a linear combination of state variables and input.
Common illustration of transfer function models and state-space models
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 41

Figure 5.17
Signal-flow graph components:
a. system;
b. signal;
c. interconnection of systems and
signals

Building Signal-Flow Charts

METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 42
 Converting Block Diagrams into Signal-Flow Charts

Figure 5.19
Signal-flow graph
development:
a. signal nodes;
b. signal-flow graph;
c. simplified signal-flow graph
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 43

Signal flow graphs of state equations

2 −5 3 x1 2
x = − 6 − 2 2 ⋅ x2 + 5 r
1 − 3 − 4 x3 7
y = −4 x1 + 6 x2 + 9 x3
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 44

Figure 5.22
Stages of development of a
signal-flow graph for the system
above:

a. place nodes;
b. interconnect state variables and
derivatives;
c. form dx1/dt ;
d. form dx2/dt
(figure continues)
 Signal flow graphs of state equations (cont'd)

2 −5 3 x1 2
x = − 6 − 2 2 ⋅ x2 + 5 r
1 − 3 − 4 x3 7
y = −4 x1 + 6 x2 + 9 x3
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 45

Figure 5.22
(continued)
e. form dx3 /dt;
f. form output

Try yourself:
Skill-Assessment Exercise 5.5 (pg 275)

Alternative representations in state space

x1 0 1 0 0 x1 0
x2 0 0 1 0 x2 0
= + u
xn −1 0 0 0 1 xn −1 0
xn − a0 − a1 − an −1 xn 1
Phase variable form:
variables are successive derivatives of each other.
Coefficients of system characteristic equation appear along the last row.
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 46

Form the signal flow graph:

Place nodes for state variables and their derivatives
Form branches connecting nodes
Other forms:
8) B 0 $ !
Cascade form
0 $
Parallel form
9) @ !
Controller canonical form
:) @ ! " 4
Observer canonical form ;) @ )
Y ( s) bn −1s n −1 + + b1 ( s ) + b0
=
U ( s ) s n + an −1s n −1 + + a1s + a0 s
 Cascade Form – Cascaded first-order systems

C (s) 24
=
R( s ) ( s + 2)( s + 3)( s + 4)

< ' !8 " $ 0 $ C 0 $ )

METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 47

−4 1 0 0
Ci ( s ) 1 x= −3 1 x+ 0 r
= 0
Ri ( s ) ( s + ai )
0 0 −2 24
y = [1 0 0]x

'$ 0
! )
@. )
4
"4 !)

Parallel Form
B ! # (
C (s) 24 12 − 24 12
= = + +
R( s ) ( s + 2)( s + 3)( s + 4) ( s + 2) ( s + 3) ( s + 4)

& "4 0" " " ! ) .)

−2 0 0 12
sX1(s) x= 0 − 3 0 x + − 24 r
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 48

0 0 −4 12
y = [1 1 1]x
sX2(s) '$ 0
#
8( " . C
.) 8 " "! !) .)
80 $ $ 0"
sX3(s) " " ) #)4) xi+=aixi+bir(
#
9( ! ) . )
4
"4 !)
 Special case parallel form – Jordan canonical form

B ! # (
C (s) ( s + 3) 2 −1 1
= = + +
R( s ) ( s + 1) 2 ( s + 2) ( s + 1) 2 ( s + 1) ( s + 2)

sX2(s) sX1(s) −1 1 0 0
x = 0 −1 0 x + 2 r
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 49

0 0 −2 1
y = [1 − 12 1]x

'$ 0 CD" !
sX3(s)
8) 3 "4
9) 3 4 "4
:) 8 !
! "4 ! "

Re-number phase variable form to get : Control(ler) Canonical Form

C (s) s 2 + 7s + 2
8) B 0 $ ! = 3
R ( s ) s + 9 s 2 + 26 s + 24
x1 0 1 0 x1 0 2. Reverse x3 0 1 0 x3 0
x2 = 0 0 1 x2 + 0 r x2 = 0 0 1 x2 + 0 r
order phase − 24 − 26 − 9 x1
x3 − 24 − 26 − 9 x3 1 variables x1 1
y = [2 7 1][x3 x1 ]
y = [2 7 1][x1 x3 ]
T
T x2
x2
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 50

:) 4 x1 − 9 − 26 − 24 x1 1
"4 x2 = 1 0 0 x2 + 0 r
" 4 x3 0 1 0 x3 0
y = [1 7 2][x1 x3 ]
T
! x2
'$ 0
!
! ! )
.)
!)
@ 0 $
"$
phase- controller ! "$
variable form canonical form
 Observer canonical form

s 2 + 7s + 2 1 s + 7 s + 2 s3
2
C (s)
= =
R( s ) s 3 + 9 s 2 + 26s + 24 1 + 9 s + 26 s 2 + 24 s 3
" $ ! ! 4
C ( s ) = 1s [R( s ) − 9C ( s )] + s12 [7 R ( s ) − 26C ( s )] + s13 [2 R ( s ) − 24C ( s )]
C ( s ) = 1s [[R ( s ) − 9C ( s )] + 1s ([7 R( s ) − 26C ( s )] + 1s [2 R ( s ) − 24C ( s )])]

−9 1 0 1
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 51

x = − 26 0 1 x + 7 r
− 24 0 0 2
y = [1 0 0]x

'$ 0
E !
! !" 8
E !
! !
E ! !
2 2 2

E = "$ ! 0 $

Feedback control system – e.g. 5.8

1. Model the forward transfer

function in cascade form (a)
2. Add feedback and input paths (b)

x1 = −3x1 + x2
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 52

x2 = − 2 x2 + 100(r − c)
y = c = 5 x1 + ( x2 − 3 x1 ) = 2 x1 + x2
x2 = −200 x1 − 102 x2 + 100r

−3 1 0
x= x+ r
− 200 − 120 100
y = [2 1]x
 Summary State Space Forms

Figure 5.31
State-space forms for

C (s) s+3
=
R( s ) ( s + 4)( s + 6)
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 53

y = c(t )

State space representations - summary

B 0 $ !
3 ! ! # :( x1 0 1 0 x1 0
x2 = 0 0 1 x2 + 0 u
Y (s) bn −1s n −1 + + b1 ( s ) + b0
= x3 − a0 − a1 − a2 x3 1
U ( s ) s n + an −1s n −1 + + a1s + a0 s
x1
y = [b0 b1 b 2 ] x2 + [0]u
x3
METR4202/7202 – Advanced Control and Robotics,Semester 2, 2006: Page 54

# ( !
'$ 0 !
x1 − a2 − a1 − a0 x1 1
x1 − a2 1 0 x1 b2
x2 = 1 0 0 x2 + 0 u
x2 = − a1 0 1 x2 + b1 u
x3 0 1 0 x3 0
x3 − a0 0 0 x3 b0
x1
x1
y = [b2 b1 b 0 ] x2 + [0]u
y = [1 0 0] x2 + [0]u
x3
x3
F G !
9 #
$( ?%" !
4 0 "4 !)