Scribd
Upload
4 views

Additional Lecture - Observability Test

This document discusses observability tests for linear time-invariant (LTI) systems, detailing the observability matrix and its significance in determining system observability. It presents various theorems, including conditions under which a system is observable based on the rank of the observability matrix, eigenvector tests, PBH tests, and Lyapunov tests. An example is provided to illustrate the application of these concepts in an electrical network system.

Uploaded by

Jahannavi Singh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
4 views

Additional Lecture - Observability Test

This document discusses observability tests for linear time-invariant (LTI) systems, detailing the observability matrix and its significance in determining system observability. It presents various theorems, including conditions under which a system is observable based on the rank of the observability matrix, eigenvector tests, PBH tests, and Lyapunov tests. An example is provided to illustrate the application of these concepts in an electrical network system.

Uploaded by

Jahannavi Singh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

Control Engineering

Ramkrishna Pasumarthy

Department of Electrical Engineering,


Indian Institute of Technology Madras
Module 12
Lecture 6

Observability Tests

Control Engineering Module 12 Lecture 6 Ramkrishna P. 1/9


Observability Tests
Observability Matrix

Consider the LTI system


/
ẋ x+ = Ax + Bu
, x ∈ Rn , u ∈ R m , y ∈ Rp (1)
y = Cx + Du
From the duality theorems it is known that the pair (A, C) is observable if and only if the
pair (A′ , C′ ) is controllable.
The controllability matrix for the pair (A′ , C′ ) is given as
[ ]
C = C′ A′ C′ (A′ )2 C′ · · · (A′ )n−1 C′
(mn)×n

The observability matrix for the pair (A, C) in (1) is defined as

O = C′
(A′ ,C′ )

Control Engineering Module 12 Lecture 6 Ramkrishna P. 2/9


Observability Matrix Test

Therefore we can write the observability matrix for the pair (A, C) as:
 
C
 CA 
 
 2 
O :=  CA  (2)

 .. 
 . 
CAn−1 (mn)×n

Theorem 8.3.1
( )
The system in (1) is observable if anf only if rank O = n.

Control Engineering Module 12 Lecture 6 Ramkrishna P. 3/9


Example

Consider the electrical network shown in figure 1 where u is the input and y is the output.

R1 R2

C1 C2
+
+ − + −
u x1 R3 x2 y

Figure 1: Electrical Network

Is it possible to develop a controller for the system where the control law is some
function of the present states x1 , x2 and the control input is updated at an interval T?

Control Engineering Module 12 Lecture 6 Ramkrishna P. 4/9


Example

The state space representation of the electrical network is


[ ] [ ]
− R11C1 0 1 [ ]
ẋ = 1 x+ u, y = 0 −1 x + R3 u
0 R2 C2 0

The observability matrix is given by


[ ]
0 −1
O=
0 − R21C2
( )
∴ rank O = 1 < 2 = n =⇒ the system is not observable.

Development of such a controller is not possible as intial states cannot be determined.

Control Engineering Module 12 Lecture 6 Ramkrishna P. 5/9


Eigenvector Test

Theorem 8.3.2

The LTI system in (1) is observable if and only if no eigenvector of A is in the kernel
of C.

Control Engineering Module 12 Lecture 6 Ramkrishna P. 6/9


PBH Test

Theorem 8.3.3

The LTI system in (1) is observable if and only if


[ ]
A − λI
rank = n, ∀ λ ∈ C
C

Control Engineering Module 12 Lecture 6 Ramkrishna P. 7/9


Lyapunov Test

Theorem 8.3.4

Assume that A is a stability matrix (Schur stable). The LTI system in (1) is observable
if and only if there is a unique positive-definite solution W to the Lyapunov equation
/
A′ W + WA = −C′ C A′ WA − W = −C′ C (3)

Moreover, the unique solution to (3) is


∫ ∞
′ ( )
W= eA τ C′ CeAτ dτ = lim WO (t0 , t1 ), continuous time case
0 (t1 −t0 ) → ∞
∞
∑ ( )
W= (A′ )k C′ CAk = lim WO (k0 , k1 ), discrete time case
(k1 −k0 ) → ∞
k=0

Control Engineering Module 12 Lecture 6 Ramkrishna P. 8/9


Overview

Summary: Module 12 Lecture 6 Contents: Module 12 Lecture 7


▶ Observability Tests. ▶ State Estimation
▶ Observer Design

Control Engineering Module 12 Lecture 6 Ramkrishna P. 9/9

You might also like