Lecture 6:

Controllability and Observability of a

Linear System

1
 Contents
 Controllability and Observability
 Definitions and Examples
 Control Gramian and Observer Gramian
 Transfer function to State Space
 Controllable canonical form

2
 Controllability & Observability
 Consider the a simple system as follows:

Can the system be influenced by two “modes” via control signal ?

Can the information about the variable be obtained by measuring output ?
 Ans: No.
 The coordinate is driven only by its initial energy or initial condition and the
control has no influence on the systemThe coordinate is not Controllable
 The output can not be used to observer the variable
 Remarks:
 The controllability and the observability concepts are associated with the
structure of the matrices , of the system
 The concepts are essential in designing controllers that have the desired effect
on the system outputs.
3
 Controllability Definition
 Consider a LTI system
(1)
 Controllability:
 A system is completely controllable if there exists a control input that can
transfer any initial state to any other desired location in a finite time
 Or Controllability of the system is formally defined as:
A state is controllable to the origin if for a given initial time there exists a finite
final time > and a piecewise continuous input signal defined on such that initial
state, the final state satisfies

4
 Controllability Definition (cont.)
 Rank test of controllability of system (1) is defined as follows:
The linear state space system (1) is controllable if and only if

 Remarks:
• It is necessary and sufficient that the matrix is invertible <= This is the
controllability criterion
 For a SISO system, is a vector, therefore, if defining
(2)
is an matrix. In this case, checking the controllability is equivalent to check
if the determinant of is nonzero.

5
 Controllability Gramian
 Consider the system

 The solution to the system at is

 Choose

 Therefore

 The value of

is called Controllability Gramian.

6
 Complete Controllability
 An system is completely controllable if any one of the following is true:
1. The controllability matrix is of full rank
2. The matrix has full row rank at every eigenvalue of
3. The controllability Gramain matrix

is non-singular and thus positive definite for every .

4. All the eigenvalues of can be assigned arbitrarily, where is an appropriately
chosen state feedback matrix and always exists.

7
 Examples
 Check the controllability of the following systems:
i)
ii) =,
iii)

Known that d is a constant.

8
 Observability Definition
 Concept of observability is dual to that of controllability, in fact
Controllability studies the possibility of steering the state from the input and
Observability studies the estimating the state from the output.
 Consider the -dimensional -input output state equation:

 Observability:
 A system is completely observable of and only if there exists a finite time
such that the initial state can be determined from the observation
history given the control input

9
 Observability Definition (cont.)
 Definition 1. The state equation is said to be observable if for any unknown
initial state , there exists a finite such that the knowledge of the input and
output over suffices to determine uniquely the initial state . Otherwise, it is
said to be unobservable.
 From the knowledge of , can be found by

 Then,

10
 Observability (i)
 Expand by using Cayley-Hamilton Theorem to obtain

 Remarks:
 The vector can only be determined if and only if the observability matrix

is invertible or in other words, the determinant of the observability matrix is nonzero.

11
 Observability Gramian
 Consider the system

 The solution

 Construct a performance index defined by

 is chosen as to minimize and it can be chosen as

where

is called Observability Gramian.

12
 Complete Observability
 A system is completely observability has the following properties
1. The observability matrix is of full rank
2. The matrix has full column rank at every eigenvalue of A
3. The Observability Gramian matrix

is non-singular and thus positive definite for every

13
 Transfer Function vs State Space
 Consider the following transfer function:

 Corresponding with the given transfer function, the following state space
equation is yielded as follows:

 Expanding the given transfer function,

which gives

where the two poles have been cancelled by two zeros.

14
 Transfer Function to State Space (cont.)
 Given that

 Controllable Canonical Form:

Let

Phase variable choice:

Phase variable form:

15
 Transfer Function to State Space (cont.)
 Because

 Then

16