Controllability and Observability

Control theory 2B-Controllability and

1
observability
Definitions
• A system is said to be controllable at time 𝑡0 if it is possible by means of an
unconstrained control vector to transfer the system from any initial state
𝑥(𝑡0 ) to any other state in a finite interval of time

Controllability deals with the possibility of forcing the system to a particular state
by application of a control input. If a state is uncontrollable then no input will be
able to control that state

• A system is said to be observable at time 𝑡0 if, with the system in state 𝑥(𝑡0 ),
it is possible to determine this state from the observation of the output over
a finite time interval

If a state is not observable then the controller will not be able to determine its
behavior from the system output and hence not be able to use that state to
stabilize the system

Controllability and observability are two properties of state models . They play an
important role in the design of control systems in state space
Control theory 2B-Controllability and
2
observability
• Although most physical systems are controllable and observable,
corresponding mathematical models may not possess the property of
controllability and observability

It is necessary to know the conditions under which

a system is controllable and observable

In what follows:

 we will derive the condition for complete state controllability

 followed by discussions of complete output controllability

 Then we will derive the condition for complete observability

 The duality principle

Control theory 2B-Controllability and

3
observability
PART 1: Complete State Controllability of Continuous-Time Systems
1.1. Case of Monovariable system

• Consider the continuous linear-time invariant system described by its state

equation
𝑥 𝑡 = 𝐴 𝑥 𝑡 + 𝐵 𝑢(𝑡) (1)

Where
𝑥 is the state vector (n-vector)
𝑢 is the control signal (scalar)  case of monovariable system
𝐴 is 𝑛 × 𝑛 matrix
𝐵 𝑖𝑠 𝑛 × 1 matrix
• The system described by Equation (1) is said to be state controllable at 𝑡 = 𝑡0 if it is
possible to construct an unconstrained control signal that will transfer an initial state
to any final state in a finite time interval t 0 ≤ 𝑡 ≤ 𝑡1

• If every state is controllable, then the system is said to be completely state

controllable
• We shall now derive the condition for complete state controllability. Without loss
of generality, we can assume that the final state is the origin of the state space and
that the initial time is zero, or 𝑡0 = 0
Control theory 2B-Controllability and
4
observability
 The condition for complete state controllability is stated as follows :

The system given by Equation (1) is completely state controllable if and only
if the vectors 𝑩, 𝑨𝑩, 𝑨𝟐 𝑩, … 𝑨𝒏−𝟏 𝑩 are linearly independent, or

𝒓𝒂𝒏𝒌 𝑩 𝑨𝑩 𝑨𝟐 𝑩 … 𝑨𝒏−𝟐 𝑩 𝑨𝒏−𝟏 𝑩 =𝒏

Kalman criterion State controllability

Control theory 2B-Controllability and

5
observability
1. 2. Case of Multivariable system

• The result just obtained can be extended to the case where the control vector 𝒖 is
𝒓-dimensional. If the system is described by
𝐁 ∈ ℝ𝒏×𝒓

𝑥 𝑡 = 𝐴 𝑥 𝑡 + 𝑩 𝑢(𝑡)

where 𝑢 is an 𝑟-vector, then it can be proved that the condition for complete state
controllability is that the 𝒏 × 𝒏𝒓 matrix

𝐵 𝐴𝐵 𝐴2 𝐵 … 𝐴𝑛−2 𝐵 𝐴𝑛−1 𝐵
be of rank 𝑛. or contain 𝑛 linearly independent column vectors

𝒓𝒂𝒏𝒌 𝑩 𝑨𝑩 𝑨𝟐 𝑩 … 𝑨𝒏−𝟐 𝑩 𝑨𝒏−𝟏 𝑩 =𝒏

The matrix
𝐵 𝐴𝐵 𝐴2 𝐵 … 𝐴𝑛−2 𝐵 𝐴𝑛−1 𝐵 : is called the Controllability Matrix

Control theory 2B-Controllability and

6
observability
Example

Consider the following systems. Are they completely state controllable. Justify your answer

System 1 System 2

The controllability Matrix is The controllability Matrix is

the system is not completely The system is

state controllable completely state controllable

Control theory 2B-Controllability and

7
observability
2. Condition for Complete State Controllability in the s-Plane (Laplace domain)

• The condition for complete state controllability can be stated in terms of

transfer functions (SISO) or transfer matrices (MIMO)

• It can be proved that a necessary and sufficient condition for complete

state controllability is that no cancellation occur in the transfer function or
transfer matrix
• If cancellation occurs, the system cannot be controlled in the direction of
the canceled mode

Example
Consider the following transfer function

𝑋(𝑠) 𝑠 + 2.5
=
𝑈(𝑠) 𝑠 + 2.5 𝑠 − 1

Clearly, cancellation of the factor (𝑠 + 2.5) occurs in the numerator and denominator
of this transfer function (Thus one degree of freedom is lost) . Because of this
cancellation, this system is not completely state controllable
Control theory 2B-Controllability and
8
observability
 3. Output Controllability

• In the practical design of a control system, we may want to control the output
rather than the state of the system
• Complete state controllability is neither necessary nor sufficient for
controlling the output of the system. For this reason, it is desirable to define
separately complete output controllability

Consider the MIMO system described by

𝑥 =𝐴𝑥+𝐵𝑢 (1.1)
𝑦 =𝐶𝑥+𝐷𝑢 (1.2)

𝑥: the state vector (n-vector, ∈ ℝ𝑛×1 )

𝑢: the control input vector (r-vector ∈ ℝ𝑟×1 )
𝑦: the output vector (m-vector ∈ ℝ𝑚×1 )

𝐴 ∈ ℝ𝑛×𝑛 ; B ∈ ℝ𝑛×𝑟 ; 𝐶 ∈ ℝ𝑚×𝑛 ; 𝐷 ∈ ℝ𝑚×𝑟

Control theory 2B-Controllability and

9
observability
• The system described by Equations (1.1)-(1.2) is said to be completely output
controllable if it is possible to construct an unconstrained control vector
𝒖 𝒕 that will transfer any given initial output 𝑦(𝑡0 ) to any final output 𝑦(𝑡1 )
in a finite time interval t 0 ≤ 𝑡 ≤ 𝑡1

The condition for complete output controllability is stated as follows :

The system described by Equations (1.1)-(1.2) is completely output controllable

if and only if the 𝒎 × 𝒏 + 𝟏 𝒓 matrix
𝐶𝐵 𝐶𝐴𝐵 𝐶𝐴2 𝐵 ⋯ 𝐶𝐴𝑛−2 𝐵 𝐶𝐴𝑛−1 𝐵 𝐷
is of rank 𝑚

𝐫𝐚𝐧𝐤 𝑪𝑩 𝑪𝑨𝑩 𝑪𝑨𝟐 𝑩 ⋯ 𝑪𝑨𝒏−𝟐 𝑩 𝑪𝑨𝒏−𝟏 𝑩 𝑫 =𝒎

Kalman criterion Output controllability

Remark
Note that the presence of the "𝑫 𝒖” term in Equation (1.2) always helps to establish
output controllability Control theory 2B-Controllability and
10
observability
4. Uncontrollable System
An uncontrollable system has a subsystem that is physically disconnected from the
input

Control theory 2B-Controllability and

11
observability
PART 2: Observability
Consider the following system
𝑥 =𝐴𝑥+𝐵𝑢
𝑦 =𝐶𝑥+𝐷𝑢

The condition for complete observability is stated as follows :

The system described by Equations (1.1) and (1.2) is completely observable if

and only if the 𝑛 × 𝑛𝑚 matrix
This matrix is called
𝐶∗ 𝐴∗ 𝐶 ∗ ⋯ (𝐴∗ )𝑛−1 𝐶 ∗
the observability matrix
is of rank 𝑛 or has 𝑛 linearly independent column vectors.

𝑤𝑖𝑡ℎ 𝐶 ∗ 𝑎𝑛𝑑 𝐴∗ stand for the conjugate transpose of the matrix C and A respectively

𝒓𝒂𝒏𝒌 𝑪∗ 𝑨∗ 𝑪∗ ⋯ 𝑨∗ 𝒏−𝟏 𝑪∗ = 𝒏
Kalman criterion for Observability

Control theory 2B-Controllability and

12
observability
Example

Consider the system described by

Is this system controllable ? Is it observable?

Control theory 2B-Controllability and

13
observability
 Solution
Since the rank of the matrix

is 2, the system is completely state controllable

For output controllability, let us find the rank of the matrix 𝐶𝐵 𝐶𝐴𝐵 𝐷
rank 𝐶𝐵 𝐶𝐴𝐵 𝐷 = 𝑟𝑎𝑛𝑘 0 1 0 = 1. Hence, the system is completely
output controllable

To test the observability condition, examine the rank of

1 1
𝐶∗ 𝐴∗ 𝐶 ∗ =
0 1

1 1
𝑟𝑎𝑛𝑘 𝐶 ∗ 𝐴∗ 𝐶 ∗ = 𝑟𝑎𝑛𝑘 = 2. Hence, the system is completely observable
0 1

Control theory 2B-Controllability and

14
observability
Conditions for Complete Observability in the s Plane
• The conditions for complete observability can also be stated in terms of transfer
functions or transfer matrices

• The necessary and sufficient conditions for complete observability is that no

cancellation occur in the transfer function or transfer matrix. If cancellation
occurs, the canceled mode cannot be observed in the output

Example
Show that the following system is not completely observable

where

Control theory 2B-Controllability and

15
observability
• Note that the control function u does not affect the complete observability of the
system. To examine complete observability, we may simply set u=0. For this
system, we have

Note that

Hence, the rank of the observability matrix is less than 3.Therefore, the system
is not completely observable

Control theory 2B-Controllability and

16
observability
In fact, in this system, cancellation occurs in the transfer function of the system. The
transfer function between 𝑋1 𝑠 𝑎𝑛𝑑 𝑈(𝑠) is

𝑋1 (𝑠) 1
=
𝑈(𝑠) 𝑠+1 𝑠+2 𝑠+3

And the transfer function between 𝑌 𝑠 𝑎𝑛𝑑 𝑋1 (𝑠) is

𝑌(𝑠)
= 𝑠+1 𝑠+4
𝑋1 (𝑠)

Therefore, the transfer function between the output is

𝑌(𝑠) 𝑌(𝑠) 𝑋1 (𝑠) (𝑠 + 1)(𝑠 + 4) Using the “zpk” Matlab
= = Instruction we get
𝑈(𝑠) 𝑋1 (𝑠) 𝑈(𝑠) (𝑠 + 1)(𝑠 + 2)(𝑠 + 3)
the same expression of
The transfer function

Clearly, the two factors 𝑠 + 1 cancel each other . This means that there are nonzero
initial states 𝑥(0) which cannot be determined from the measurement of 𝑦(𝑡)

Control theory 2B-Controllability and

17
observability
Remark

The transfer function has no cancellation if and only if the system is completely
state controllable and completely observable

This means that the canceled transfer function does not carry along all the
information characterizing the dynamic system

Control theory 2B-Controllability and

18
observability
Part 3: Principle of Duality
• We shall now discuss the relationship between controllability and observability

• We shall introduce the principle of duality, due to Kalman, to clarify apparent

analogies between controllability and observability

• Consider the systems 𝑺𝟏 and S2 described by

System S1 System S2

Where Where

Control theory 2B-Controllability and

19
observability
 The principle of duality states that the system S1 is completely state controllable
(observable) if and only if system S2 is completely observable (state controllable)

To verify this principle, let us write down the necessary and sufficient conditions
for complete state controllability and complete observability for systems S1 and
S2

System S1 System S2

A necessary and sufficient condition for A necessary and sufficient condition for
complete state controllability is that complete state controllability is that
𝒓𝒂𝒏𝒌 𝑩 𝑨𝑩 𝑨𝟐 𝑩 … 𝑨𝒏−𝟏 𝑩 = 𝒏
𝒓𝒂𝒏𝒌 𝑪∗ 𝑨∗ 𝑪∗ ⋯ 𝑨∗ 𝒏−𝟏 𝑪∗ =𝒏

A necessary and sufficient condition for A necessary and sufficient condition for
complete observability is that complete observability is that the

𝒓𝒂𝒏𝒌 𝑩 𝑨𝑩 𝑨𝟐 𝑩 … 𝑨𝒏−𝟏 𝑩 =𝒏
𝒓𝒂𝒏𝒌 𝑪∗ 𝑨∗ 𝑪∗ ⋯ 𝑨 ∗ 𝒏−𝟏 ∗
𝑪 =𝒏

Control theory 2B-Controllability and

20
observability
Comments

• By comparing these conditions, the truth of this principle is apparent

• By use of this principle, the observability of a given system can be checked by

testing the state controllability of its dual

Control theory 2B-Controllability and

21
observability
Exercise
A state-space representation of a system in the controllable canonical form is given by

The same system may be represented by the following state-space equation, which is
in the observable canonical form

1- Show that the state-space representation given by the Controllable Canonical

Form gives a system that is completely state controllable, but not observable
2- Show, on the other hand, that the state-space representation defined by
Observable Canonical Form gives a system that is not completely state controllable,
but is observable

3- Explain what causes the apparent difference in the controllability

and observability of the same system
Control theory 2B-Controllability and
22
observability
Solution

1-Consider the system defined by Equations the Controllable CF, The rank of the
controllability matrix is 2. Hence, the system is completely state controllable

The rank of the observability matrix is 1. Hence the system is not observable

2- Consider the system defined by the Observable CF, The rank of the controllability
matrix is 1. Hence, the system is not completely state controllable

The rank of the observability matrix is 2. Hence, the system is observable

Control theory 2B-Controllability and

23
observability
3- The apparent difference in the controllability and observability of the same system
is caused by the fact that the original system has a pole-zero cancellation in the
transfer function

• Using the correlation between SS and TF (case where the matrix D=0) then we
have
𝑮 𝒔 = 𝑪 𝒔 𝑰 − 𝑨 −𝟏 𝑩 + 𝑫
• If we use the Controllable CF to compute the expression of G(s) (Note that the same
transfer function can be obtained by using the Observable CF), then

Using the “zpk”

Matlab
Instruction
we get
the same
expression of the
transfer function

Clearly, cancellation occurs

in this transfer function
Control theory 2B-Controllability and
24
observability
 The considered system in this example is represented
by a State Space Representation (SSR )

Its Controllable Canonical Form Its Observable canonical form

The system in the Controllable CF is The system in the Observable CF is

completely state controllable not completely state controllable but
but not observable is observable

The same system (𝒔 + 𝟎. 𝟖)

𝑮 𝒔 =
is represented by its Transfer function (𝒔 + 𝟎. 𝟖)(𝒔 + 𝟎. 𝟓)

• If a pole-zero cancellation occurs in the transfer function, then the controllability

and observability vary, depending on how the state variables are chosen
• Remember that, to be completely state controllable and observable, the
transfer function must not have any pole-zero cancellations
Control theory 2B-Controllability and
25
observability