Lecture 10

Conversion Between State Space and

Transfer Function Representations in
Linear Systems II

Dr. Radhakant Padhi

Asst. Professor
Dept. of Aerospace Engineering
Indian Institute of Science - Bangalore
 An Alternate First Companion Form
(Toeplitz first companion form)

b n s n + b n 1 s n 1 + + b1 s + b 0 y(s)
H (s) = n n 1 =
s + a n 1 s + + a1 s + a 0 u ( s )
i .e .
y ( n ) + a n 1 y ( n 1) + + a1 y + a 0 y
+ b n 1u (
n 1)
+ bn u (
n)
= b 0 u + b1 u +

Ref: B. Friedland,
Control System Design
Mc Graw Hill, 1986.

ADVANCED CONTROL SYSTEM DESIGN 2

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
An Alternate First Companion Form
(Toeplitz first companion form)

Define the state variables x1 , , xn such that

y = x1 + p0u
x1 = x2 + p1u
x2 = x3 + p2u

xn 1 = xn + pn 1u
xn = an 1 xn a0 x1 + pnu
ADVANCED CONTROL SYSTEM DESIGN 3
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
An Alternate First Companion Form
(Toeplitz first companion form)

From the above definition, we have

y = x1 + p 0 u
y = x1 + p 0 u = ( x 2 + p 1 u ) + p 0 u
y = x 2 + p1u + p 0 u = ( x 3 + p 2 u ) + p1u + p 0 u

( n 1) (n 2 ) ( n 1)
y = x n + p n 1u + p n 2 u + + p1u + p0u
y ( n ) = a n 1 x n a n 2 x n 1 a 0 x1 + p n u
+ p1u (
n 1)
+ p0u (
n)
+ p n 1u + p n 2 u +
ADVANCED CONTROL SYSTEM DESIGN 4
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
An Alternate First Companion Form
(Toeplitz first companion form)
(n ) ( n 1)
y + a n 1 y + + a1 y + a 0 y
= ( p n + a n 1 p n 1 + + a1 p1 + a 0 p 0 ) u
+ ( p n 1 + + a 2 p1 + a1 p 0 ) u
+
+ ( p1 + a n 1 p 0 ) u ( n 1)

(n )
+ p0u
= b 0 u + b1 u + + b n 1u ( n 1) + b n u ( n )
(fro m th e T F )
ADVANCED CONTROL SYSTEM DESIGN 5
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
An Alternate First Companion Form
(Toeplitz first companion form)

Equating the coefficients:

p0 = bn
p1 + an 1 p0 = bn 1

pn 1 + an 1 pn 1 + + a1 p0 = b1
pn + an 1 pn 1 + + a0 p0 = b0

Now, we need to solve for ( p0 , p1 , , pn )

ADVANCED CONTROL SYSTEM DESIGN 6

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
An Alternate First Companion Form
(Toeplitz first companion form)
In matrix form:
1 0 0 0 p0 bn
a 1 0 0 p1 bn 1
n 1
an 2 an 1 0 =


a0 a1 an 1 1 pn b0
Toeplitz Matrix

Note : Toeplitz matrix is a nonsinular matrix

( determinant = 1 0 )
Hence, the solution always exists!
ADVANCED CONTROL SYSTEM DESIGN 7
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
 An Alternate First Companion Form
(Toeplitz first companion form)

State space form (from definition of state & output variables):

x1 0 1 0 0 x1 p0
x 0 0 1 0 x2 p1
2 = + u


xn a0 a1 an 1 xn pn
y = [1 0 0] X + [ p0 ] u

Note : The physical meaning of state variables

in different companion froms are differnet.

ADVANCED CONTROL SYSTEM DESIGN 8

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
Block Diagram for realization of
Toeplitz first companion form

ADVANCED CONTROL SYSTEM DESIGN 9

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
Alternate/Toeplitz first companion
form: Some comments
z Toeplitz realization also requires
n integrators
z Extension of Toeplitz realization is
straightforward to MISO systems
One need to solve m-system of linear
equations, where m is the number of
input (control) variables.
However, only n integrators are
needed.

ADVANCED CONTROL SYSTEM DESIGN 10

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
 Second Companion Form
(Observable Canonical Form)

bn s n + bn 1s n 1 + + b1s + b0 y ( s )
H (s) = n n 1 =
s + an 1s + + a1s + a0 u ( s )
i.e.
(s n
+ an 1s n 1 + + a1s + a0 ) y ( s )
= ( bn s n + bn 1s n 1 + + b1s + b0 ) u ( s )
Rearranging the terms:
s n [ y ( s ) bnu ( s ) ] + s n 1 [ an 1 y ( s ) bn 1u ( s ) ] +
+ [ a0 y ( s) b0u ( s) ] = 0
ADVANCED CONTROL SYSTEM DESIGN 11
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
 Second Companion Form
(Observable Canonical Form)
Simplify:
1 1
[ y( s) bnu ( s)] = [bn1u (s) an1 y (s)] + + n [b0u ( s ) a0 y ( s ) ]
s s
Solve for y(s):
1 1
y ( s) = bn u ( s) + [ n1
b u ( s ) an 1 y ( s ) ] + + n [ 0
b u ( s ) a0 y ( s ) ]
s s

1 1
= bnu ( s ) + [bn 1u ( s ) an 1 y ( s ) ] + [bn 2u ( s ) an 2 y ( s ) ] +
s s
x2 ( s )
x1 ( s )

ADVANCED CONTROL SYSTEM DESIGN 12

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
Block Diagram Realization

ADVANCED CONTROL SYSTEM DESIGN 13

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
 Second Companion Form
(Observable Canonical Form)

The equations can be written as:

y = x1 + bnu
x1 = ( x2 an 1 y + bn 1u ) = an 1 x1 + x2 + ( bn 1 an 1bn ) u
x2 = ( x3 an 2 y + bn 2u ) = an 2 x1 + x3 + ( bn 2 an 2bn ) u

xn 1 = ( xn a1 y + b1u ) = a1 x1 + xn + ( b1 a1bn ) u
xn = b0u a0 y = a0 x1 + ( b0 a0bn ) u

ADVANCED CONTROL SYSTEM DESIGN 14

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
Second Companion Form
(Observable Canonical Form)
In vector-matrix form, we can write it as
x1 an 1 1 0 0 0 x1 bn 1 an 1bn
x a 0 1 0 0 x2 bn 2 an 2bn
2 n2
= + u

n 1 a1
x 0 0 0 1 xn 1 b1 a1bn
xn a0 0 0 0 0 xn b0 a0bn
X A X B

y = [1 0 0] X + [bn ] u
C D

ADVANCED CONTROL SYSTEM DESIGN 15

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
Second Companion Form
(Observable Canonical Form)
(1) In second companion form, the coefficients of the
denominator of the transfer function appear in a
column, whereas in the first companion form they
appear in a row.
(2) Extensions of second companion form to SIMO,
MISO cases etc. are possible (but beyond the scope
this course)
(3) Controllable canonical form is good for control
design, whereas observable canonical form is good
for observer/filter design
ADVANCED CONTROL SYSTEM DESIGN 16
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
Comment on minimal realization
(for MIMO systems)
We have seen realization of:
SIMO systems: n integrators
MISO systems: n integrators
Question: How about MIMO systems? Will it still be possible to
realize it with n integrators?
Answer: No!
However, one way of realizing MIMO systems will be to use a
number of structures (of either SIMO or MISO form) in parallel; i.e.
if U m and Y p
then it is always possible to realize such a MIMO system with no
more than n x min (m, p) integrators.

ADVANCED CONTROL SYSTEM DESIGN 17

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
Comment on minimal realization
(for MIMO systems)
Question: How about fewer integrators?
Answer: This leads to the questions of minimal realization;
a subject of considerable research during 1970s.
Why necessary? Because a non-minimal realization is
either non-controllable or non-observable (or both). It may
cause theoretical and computational problems too.
Solution: Several algorithms exist for finding a minimal
realization and corresponding A,B,C,D matrices. However,
these are beyond the scope of this course!
Further reading: T. Kailath, Linear Systems, Prentice Hall, 1980.

ADVANCED CONTROL SYSTEM DESIGN 18

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
ADVANCED CONTROL SYSTEM DESIGN 19
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
ADVANCED CONTROL SYSTEM DESIGN 20
Dr. Radhakant Padhi, AE Dept., IISc-Bangalore