Scribd
Upload
151 views17 pages

The Controller Canonical Form: Topics

This document discusses the controller canonical form (CCF) for representing linear systems. The key points are: - The CCF uses a linear combination of the columns of the controllability matrix as a basis to represent the state space. - It allows arbitrarily placing the closed-loop poles using full-state feedback if the system is completely controllable. - The procedure is to first transform the system into CCF, then place the desired eigenvalues using state feedback, and finally transform back to find the control gains for the original system. - An example demonstrates transforming a system to CCF, placing the closed loop poles, and finding the feedback gains to achieve this placement for the original system.

Uploaded by

Chandan Kumar
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
151 views17 pages

The Controller Canonical Form: Topics

This document discusses the controller canonical form (CCF) for representing linear systems. The key points are: - The CCF uses a linear combination of the columns of the controllability matrix as a basis to represent the state space. - It allows arbitrarily placing the closed-loop poles using full-state feedback if the system is completely controllable. - The procedure is to first transform the system into CCF, then place the desired eigenvalues using state feedback, and finally transform back to find the control gains for the original system. - An example demonstrates transforming a system to CCF, placing the closed loop poles, and finding the feedback gains to achieve this placement for the original system.

Uploaded by

Chandan Kumar
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 17

18.

The Controller Canonical Form


Topics
Be able to

obtain a state variable representation using the controllability matrix (this is the controller canonical form).

.
.

derive the controller canonical form.


show that the closed -loop poles can be arbitrarily placed using full-state feedback if the system is completely controllable. design a full-state feedback controller using the controller canonical form.
Fall 200.

Ref.: Sections 6.3-6.5.

Canonical Forms
If an n-th order linear system is completely
controllable, then [B AB A2B... An-IB] has n linearly independent columns.

These columns (or their linear combination) can be used as basis vectors to obtain canonical state-space representations.
This can also be done with the observability test matrix.

The Controller Canonical Form .


use a linear combination of the columns of the controllability test matrix as the basis for the state space. 1 0 ... 0 0
al a2

1
al

... ...

0 0 0 0

p = [An-l B : An-2B ... AD : B]'


an-2 an-l an-3 an-2

...

1 0
1

... al

and the ai's are from the characteristic equation, Ll ( A) = An + a 1 A n-l + a2 !\In-2 + ... + a n-l A + a n

controller canonical form (cant):

What is the state variable model (Ac, Bc, Cc) using this basis? Find Ac when xCI)= Pxc(t), ic(t) =
p-l APxc(t) + p-l Hu(t)

Find Ac from AP = PAc.

Consider the case where n = 3, A[A2B + alAn + a2B : AB + aID: B] =

[A2B+ alAn + a2B : AB + aIB : B] Ac Find a C3the third column of A C, ,

controller canonical form (cont):

Find the second column of A c '

Find the first column of A . (Dse the CayleyHamilton theorem here.) c

Final result for n = 3, 0 1 0 A c =10


0 1
-a 3 -a 2 -a 1

controller canonical form (cont):

Find Be from p-l B =B c ,

B = [A2B + a1AB + a2B : AB + a1B : B] B c

Find C c ,

Now generalize

controller canonical form (cont):

Given, b s n-I + b s n-2+ ... + b

s+b
n -I n

G(s)

S n + a 1 s n-I + a 2 s n-2 +

... + an-I s + an

the controller canonical form (CCF) is


0 0
1

0
1

. . . . . .

0
'0

0
0

AC

. .
0
0

. .
.

'.

. . .
0
0 -a n-2

. . .

. . .
0

0
0

. . . . . .
. . .

0
-a
2

1
-a
1

-a n -a n-l

Be = [0 0
Ce = [bn bn-I

0 l]T
b2 bI]

Note: This form can be obtained directly from the transfer function.
The controller canonical realization

:ill

XII e

:i1l-1 e

XII-Ie

nG.""1

a..1iad9n

of me conuoIIer canonial form.

Verify

the CCF

Compute the closed-loop transfer function Ce(sI

- Ae)-l Be

First,
s -1

. .
.

. .

. .

0 0 0

0 0 0

0 0
I
.

s 0

-1 s

.
.

(sI-A)

= I.
.

.
.

. .
.
. .
.

.
.

. 0 an-Z

.
-1

an an-l

. . .

az s +al

Since Be = [0 0 ... 0 1]T, (sI-Ae)-1 Be will be the

last column (LC) of (sI-Ae)-l

verification (cont)..

(LC)(sI _A)-l

= (LC) Adj(sI -A) det(s I - A c)

of (sI - Ac) for n = 3 s


sI
-

(LC) Adj(sI

- Ac) = cofactors of the last row

-1

0 -1
~

A c =10

(LC)Adj (sI - Ac)

3 2 S+OI

In general, (LC) Adj (sI - Ac) = (1 S S2 ... sn-l)T . det (sI - A c) = sn + a 1sn-l+ a2 sn-2+... -n-l-n s + ~ ~

verification (cont):

multiplying by Cc yields,
1

s
S2

(bn bn-I

... b2 bl)

.
.
.

det(sI -Ac)

S n-I

b +b
n n-I

s + . . . + b s n-2 + b s n-I
2 I

s n + a I s n-I +a 2s n-2+ . . . + an-l s + a n

G(s)

Alternate Derivation of the CCF


b 1s+b2

Consider G(s) = s2+a1s+a2 Multiply G(s) by s-2F(s)/S-2F(s),then yes) U(s)

Equate the numerators and denominators:

yes) = U(s) = Solve (*) for F(s)


F(.)

(*)
b1

Xz(S)

X\(s)

1/s

Y(s)

-~

Effect of Full-State Feedback


Introduceu(t) -K x c (I) to the previous block
==
C

diagrann

Xn e

Xn e

in-Ie

Xn-le

...
c -kn-t -a2
~ -k2-an-l

Xl e

Xl e

c -kn-at

c
.

-hI-an

FIG. 6.5-1

The effect of full-state feedback in the controller canonical form.

full-state feedback (cont)..

How can you shift the original Ab "" " AI' A2' ... An? ~(s) = (s-i1)(s-i2)

A2

...

A n to

...(s-in)

or
S n +

a 1s n-l

a2s n-2

...

an

Using the preceding block diagram find


k1 , k2c, ..., knc, C

Procedure
1. Given (A, B, C), transform it by P to
(Ac,Bc' C c) . The eigenvalues remain the same.

2. In controller canonical form, move the eigenvalues by using


u(t)
= -

K c x c (t)

3. Transform back to find the control gains for (A, B, C):


u(t) = -KCxc(t)
=

-Kcp-lX(t)

Example
For the open-loop system with

A=-12 0

B=

1
1

C =[1 0]

place the closed-loop poles at -3 and -4.


Ans. k)

= -3, k2= 12

Solution (A, B) is completely controllable.


1. Find ~(A) to obtain the a/s ,

Find al and a2 from ~(A),

2. Shift the poles of A c to -3 and -4.

3. What are the control gains for (A, B, C)?

You might also like