Anda di halaman 1dari 23

ELEC4410

Control Systems Design


Lecture 16: Controllability and Observability
Canonical Decompositions

Julio H. Braslavsky
julio@ee.newcastle.edu.au

School of Electrical Engineering and Computer Science


The University of Newcastle

The University of Newcastle Lecture 15: Observability p.1/14


Outline
Canonical Decompositions
Kalman Decomposition and Minimal Realisation
Discrete-Time Systems

The University of Newcastle Lecture 15: Observability p.2/14


Canonical Decompositions
The Canonical Decompositions of state equations will establish
the relationship between Controllability, Observability , and a
transfer matrix and its minimal realisations.

Consider the state equation

x = Ax + Bu A Rnn , B Rnp ,
where pn qp
(SE)
y = Cx + Du CR ,D R .

Let x = Px, where P is nonsingular, P Rnn . Then we know that


the state equation

x = Ax + Bu A = PAP1 , B = PB,
where
y = Cx + Du C = CP1 , D = D,

is algebraically equivalent to (SE).

The University of Newcastle Lecture 15: Observability p.3/14


Canonical Decompositions
Theorem (Controllable/Uncontrollable Decomposition). Consider the
n-dimensional state equation (SE) and suppose that
h i
rank C = rank B AB An1 B = n1 < n

(i.e., the system is not controllable). Let the n n matrix of change of


coordinates P be defined as
h i
1
P = q 1 q2 q n1 q n

where the first n1 columns are any n1 independent columns in C, and


the remaining are arbitrarily chosen so that P is nonsingular. Then the
equivalence transformation x = Px transforms (SE) to

x C A A12 x B
= C C + C u
x Ce 0 ACe xCe 0
h i
y = CC CCe x + Du

The University of Newcastle Lecture 15: Observability p.4/14


Canonical Decompositions
Controllable
The states in the new coordinates are u y
decomposed into C

xC : n1 controllable states
e
C
xCe : n n1 uncontrollable states
Uncontrollable

The University of Newcastle Lecture 15: Observability p.5/14


Canonical Decompositions
Controllable
The states in the new coordinates are u y
decomposed into C

xC : n1 controllable states
e
C
xCe : n n1 uncontrollable states
Uncontrollable

The reduced order state equation of the controllable states

x C = AC xC + BC u
y = CC x + Du

is controllable and has the same transfer function as the original state
equation (SE).

The M ATLAB function ctrbf transforms a state equation into its control-
lable/uncontrollable canonical form.

The University of Newcastle Lecture 15: Observability p.5/14


Canonical Decompositions
Theorem (Observable/Unobservable Decomposition). Consider the
n-dimensional state equation (SE) and suppose that
 C 
rank O = rank CA
= n2 < n (i.e., the system is not observable).
CAn1

Let the n n matrix of change of coordinates P be defined as


p1
p2
P= p
n2


pn

where the first n2 columns are any n2 independent columns in O, and


the remaining are arbitrarily chosen so that P is nonsingular. Then the
equivalence transformation x = Px transforms (SE) to

x A 0 x B
O = O O + O u
x O
e A21 AO
e xOe BOe
h i
y = CO 0 x + Du
The University of Newcastle Lecture 15: Observability p.6/14
Canonical Decompositions
The states in the new coordinates are
e
O
decomposed into
Unobservable
xO : n2 observable states u y
O
e :
xO n n2 unobservable states
Observable

The University of Newcastle Lecture 15: Observability p.7/14


Canonical Decompositions
The states in the new coordinates are
e
O
decomposed into
Unobservable
xO : n2 observable states u y
O
e :
xO n n2 unobservable states
Observable

The reduced order state equation of the observable states

x O = AO xO + BO u
y = CO x + Du

is observable and has the same transfer function as the original state
equation (SE).

The M ATLAB function obsvf transforms a state equation into its observ-
able/unobservable canonical form.

The University of Newcastle Lecture 15: Observability p.7/14


Kalman Decomposition
The Kalman decomposition combines the
Controllable/Uncontrollable and Observable/Unobservable
decompositions.
Every state-space equation can be
e
CO transformed, by equivalence transfor-
mation, into a canonical form that splits
Unobservable
the states into
u y
CO Controllable and observable states

e
CO

eO
C e

Uncontrollable

The University of Newcastle Lecture 15: Observability p.8/14


Kalman Decomposition
The Kalman decomposition combines the
Controllable/Uncontrollable and Observable/Unobservable
decompositions.

Every state-space equation can be


e
CO
transformed, by equivalence transfor-
Unobservable mation, into a canonical form that splits
u y the states into
CO
Controllable and observable states
Controllable but unobservable
e
CO states

eO
C e

Uncontrollable

The University of Newcastle Lecture 15: Observability p.8/14


Kalman Decomposition
The Kalman decomposition combines the
Controllable/Uncontrollable and Observable/Unobservable
decompositions.

Every state-space equation can be


e
CO
transformed, by equivalence transfor-
Unobservable mation, into a canonical form that splits
u y the states into
CO
Controllable and observable states
Controllable but unobservable
e
CO states
Uncontrollable but observable
states
eO
C e

Uncontrollable

The University of Newcastle Lecture 15: Observability p.8/14


Kalman Decomposition
The Kalman decomposition combines the
Controllable/Uncontrollable and Observable/Unobservable
decompositions.

Every state-space equation can be


e
CO
transformed, by equivalence transfor-
Unobservable mation, into a canonical form that splits
u y the states into
CO
Controllable and observable states
Controllable but unobservable
e
CO states
Uncontrollable but observable
states
eO
C e
Uncontrollable and unobservable
Uncontrollable states

The University of Newcastle Lecture 15: Observability p.8/14


Kalman Decomposition
The Kalman decomposition brings the system to the form

x CO ACO 0 A13 0 xCO BCO

x A
CO e 21 ACOe A23 A24 xCO e BCO e
= + u
x e 0 0 ACO 0 e 0
CO e xCO
x C
eOe 0 0 A43 AC
eOe xC
eOe 0
| {z }
x
h i
y = CCO 0 CCO
e 0 x + Du

The University of Newcastle Lecture 15: Observability p.9/14


Kalman Decomposition
The Kalman decomposition brings the system to the form

x CO ACO 0 A13 0 xCO BCO

x A
CO e 21 ACOe A23 A24 xCO e BCO e
= + u
x e 0 0 ACO 0 e 0
CO e xCO
x C
eOe 0 0 A43 AC
eOe xC
eOe 0
| {z }
x
h i
y = CCO 0 CCO
e 0 x + Du

A minimal realisation of the system is obtained by using only the


controllable and observable states from the Kalman
decomposition.

x CO = ACO xCO + BCO u


y = CCO x + Du
The University of Newcastle Lecture 15: Observability p.9/14
Kalman Decomposition
Example. Consider the system in Modal Canonical Form
" #  
1 1 0 0 1
0 1 0 0 0
x = 0 0 2 0 x+ 0 u
0 0 0 3 2

y = [1 0 1 1]x

From the example seen in the Tutorial, Controllability and


Observability in Modal Form equations, we see that
the first 1 is controllable and observable
2 is not controllable, although observable
3 is controllable and observable
Thus a minimal realisation of this system is given by
h i 1
x = 1 0
0 3 x + 2 u 1 2
with transfer function G(s) = +
y = [ 1 1 ] x s 1 s 3
The University of Newcastle Lecture 15: Observability p.10/14
Canonical Decompositions
Example (Controllable/Uncontrollable decomposition).
Consider the third order system
h1 1 0i h0 1i
x = 0 1 0 x + 1 0 u
011 01
y = [1 1 1]x

Compute the rank of the controllability matrix,


h0 1 1 1 2 1i
rank C = rank [ B AB A2 B ] = rank 1 0 1 0 1 0 = 2 < 3,
011121

thus the system is not controllable. Take the change of


coordinates formed by the first two columns of C and an
arbitrary third one independent of the first two,
h0 1 1i
P 1
=Q, 100 .
010

The University of Newcastle Lecture 15: Observability p.11/14


Canonical Decompositions
Example (continuation). By doing ^
x = Px we obtain the equivalent
equations

..
1 0 . 0  1 0 
..
x =
^ 1 1 .

0 x +
0 1
... ... u
... ... ... ... 0 0
..
0 0 . 1
 
.
y = 1 2 .. 1 x

and the reduced controllable system

x = [ 11 01 ] x + [ 10 01 ] u
^
y = [1 2]x

which has the same transfer matrix than the original system
h i
s+1
G(s) = s2 2s+1 s1 .2

The University of Newcastle Lecture 15: Observability p.12/14


Discrete-Time Systems
For controllability and observability of a discrete-time equation

x[k + 1] = Ax[k] + Bu[k]


y[k] = Cx[k] + Du[k]

we can use the same Controllability and Observability matrices


rank tests that we have for continuous-time systems,

rank C = rank [ B AB An1 B ] = n Controllability


 C 
rank O = rank CA
= n Observability
CAn1

Canonical decompositions are analogous.

The University of Newcastle Lecture 15: Observability p.13/14


Summary
When a system is not controllable or not observable, there
might be a part of the system that still is controllable and
observable.

The University of Newcastle Lecture 15: Observability p.14/14


Summary
When a system is not controllable or not observable, there
might be a part of the system that still is controllable and
observable.
The controllability and observability matrices can be used to
split (by a change of coordinates) a state equation into its
controllable/uncontrollable parts and
observable/unobservable parts.

The University of Newcastle Lecture 15: Observability p.14/14


Summary
When a system is not controllable or not observable, there
might be a part of the system that still is controllable and
observable.
The controllability and observability matrices can be used to
split (by a change of coordinates) a state equation into its
controllable/uncontrollable parts and
observable/unobservable parts.
The controllable and observable part of a state equation
yields minimal realisation .

The University of Newcastle Lecture 15: Observability p.14/14


Summary
When a system is not controllable or not observable, there
might be a part of the system that still is controllable and
observable.
The controllability and observability matrices can be used to
split (by a change of coordinates) a state equation into its
controllable/uncontrollable parts and
observable/unobservable parts.
The controllable and observable part of a state equation
yields minimal realisation .

Thus, we conclude that for a state equation


minimal realisation controllable and observable

The University of Newcastle Lecture 15: Observability p.14/14

Anda mungkin juga menyukai