0% found this document useful (0 votes)
38 views33 pages

3 Controllabillity&observability

The document discusses concepts of controllability and observability for control systems. It defines controllability as the ability to transfer a system between any two states using control inputs, and observability as the ability to determine the system state from output measurements. Conditions for complete controllability and observability are presented, including that the controllability and observability matrices must have full rank.

Uploaded by

sasamagdy1512
Copyright
© © All Rights Reserved
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
0% found this document useful (0 votes)
38 views33 pages

3 Controllabillity&observability

The document discusses concepts of controllability and observability for control systems. It defines controllability as the ability to transfer a system between any two states using control inputs, and observability as the ability to determine the system state from output measurements. Conditions for complete controllability and observability are presented, including that the controllability and observability matrices must have full rank.

Uploaded by

sasamagdy1512
Copyright
© © All Rights Reserved
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/ 33

CONCEPTS OF CONTROLLABILITY AND OBSERVABILITY

• A SYSTEM IS SAID TO BE CONTROLLABLE AT TIME 𝒕𝒐 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 TIME INTERVAL.
• A SYSTEM IS SAID TO BE OBSERVABLE AT TIME 𝒕𝒐 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.
CONCEPTS OF CONTROLLABILITY AND OBSERVABILITY

• THE CONCEPTS OF CONTROLLABILITY AND OBSERVABILITY WERE


FIRST INTRODUCED BY KALMAN

• THE CONDITIONS OF CONTROLLABILITY AND OBSERVABILITY MAY


GOVERN THE EXISTENCE OF A COMPLETE SOLUTION TO THE CONTROL
SYSTEM DESIGN PROBLEM.

• IT IS NECESSARY TO KNOW THE CONDITIONS UNDER WHICH A


SYSTEM IS CONTROLLABLE AND OBSERVABLE.
SOLUTION OF NONHOMOGENEOUS EQUATIONS

𝒙 𝒕 = 𝑨 𝒙 𝒕 + 𝑩 U(t)
𝒙 𝒕 − 𝑨 𝒙 𝒕 = 𝑩 U(t)
Multiplying both sides of this equation by 𝒆−𝑨𝒕
𝒅 −𝑨𝒕
𝒆−𝑨𝒕 𝒙 𝒕 −𝑨𝒙 𝒕 = 𝒆 𝒙 𝒕 = 𝒆−𝑨𝒕 𝑩 U(t)
𝒅𝒕
Integrating between 0 & t gives
𝒕 −𝑨𝝉
𝒆−𝑨𝒕 𝒙 𝒕 − 𝒙 𝟎 = 𝟎
𝒆 𝑩 U(𝝉) 𝒅𝝉

Or multiplying both sides by 𝒆𝑨𝒕 yields


𝒕 −𝑨𝝉 𝒕 𝑨(𝒕−𝝉)
𝒙 𝒕 = 𝒆𝑨𝒕 𝒙 𝟎 + 𝒆𝑨𝒕 𝟎
𝒆 𝑩 U(𝝉) 𝒅𝝉 = 𝒆𝑨𝒕 𝒙 𝟎 + 𝟎
𝒆 𝑩 U(𝝉) 𝒅𝝉
COMPLETE STATE CONTROLLABILITY OF CONTINUOUS TIME SYSTEMS
𝒙 𝒕 = 𝑨 𝒙 𝒕 + 𝑩 U(t)
𝒕
𝒙 𝒕 = 𝒆𝑨𝒕 𝒙 𝒐 + 𝒆𝑨(𝒕−𝝉) 𝑩 𝑼 𝝉 𝒅𝝉
𝟎

Assume that the final state is the origin of the state space and that the initial time is zero, or 𝒕𝒐 = 𝟎.
𝒕𝒇
𝑨𝒕𝒇
𝒙 𝒕𝒇 = 𝒆 𝒙 𝒐 + 𝒆𝑨(𝒕𝒇−𝝉) 𝑩 𝑼 𝝉 𝒅𝝉
𝟎

or
𝒕𝒇
𝒙 𝒐 =− 𝒆−𝑨𝝉 𝑩 𝑼 𝝉 𝒅𝝉
𝟎
𝒏−𝟏
𝒆−𝑨𝝉 = 𝜶𝒌 𝝉 𝑨𝒌
𝒌=𝟎
𝒏−𝟏 𝒕𝒇
𝒌
𝒙 𝒐 =− 𝑨 𝑩 ∝𝒌 (𝝉) 𝑼 𝝉 𝒅𝝉
𝒌=𝟎 𝟎
COMPLETE STATE CONTROLLABILITY OF CONTINUOUS TIME SYSTEMS

LE T
𝑡𝑓
𝛽𝑘 = 0
∝𝑘 (𝜏) 𝑈 𝜏 𝑑𝜏
𝑛−1 𝑘
𝑥 𝑜 =− 𝑘=0 𝐴 𝐵 𝛽𝑘

𝛽𝑜
𝛽1
= − 𝐵: 𝐴𝐵: … . . 𝐴 𝐵 … . .
𝑛−1
………
𝛽𝑛−1

FOR THE PREVIOUS EQUATION TO BE SATISFIED

THE RANK OF THE nxn matrix 𝐵: 𝐴𝐵: … . . 𝐴𝑛−1 𝐵 SHOULD BE n


CONDITIONS FOR COMPLETE CONTROLLABILITY

The system is completely state controllable if and only if the


vectors 𝑩: 𝑨𝑩: … . . 𝑨𝒏−𝟏 𝑩 are linearly independent or the nxn
matrix 𝑩: 𝑨𝑩: … . . 𝑨𝒏−𝟏 𝑩 has a rank n

In case U is an r dimensional vector then the nxnr matrix


𝑩: 𝑨𝑩: … . . 𝑨𝒏−𝟏 𝑩 is of rank n
The matrix 𝑩: 𝑨𝑩: … . . 𝑨𝒏−𝟏 𝑩 is the controllability matrix.
EXAMPLE: I

CONSIDER THE SYSTEM GIVEN BY:-

𝒙𝟏 𝟏 𝟏 𝒙𝟏 𝟏
= 𝒙 + U
𝒙𝟐 𝟎 −𝟏 𝟐 𝟎
Since
𝟏 𝟏
𝑩 ⋮ 𝑨𝑩 = = 𝒔𝒊𝒏𝒈𝒖𝒍𝒂𝒓
𝟎 𝟎

∴ The system is not completely state controllable


EXAMPLE: II

CONSIDER THE SYSTEM GIVEN BY:-

𝒙𝟏 𝟏 𝟏 𝒙𝟏 𝟎
= 𝒙 + U
𝒙𝟐 𝟐 −𝟏 𝟐 𝟏
Since
𝟎 𝟏
𝑩 ⋮ 𝑨𝑩 = = 𝒏𝒐𝒏𝒔𝒊𝒏𝒈𝒖𝒍𝒂𝒓
𝟏 −𝟏
∴ The system is therefore completely state controllable
ALTERNATE FORM OF THE CONDITION FOR COMPLETE STATE CONTROLLABILITY

Consider the system 𝒙 𝒕 = 𝑨 𝒙 𝒕 + 𝑩 U(t)


if the eigenvalues of A are distinct, then it is possible to find a transformation matrix
P such that:

𝝀𝟏 … … 𝟎
⋮ 𝝀𝟐 … …
𝑷−𝟏 𝑨𝑷 = 𝑫 =
⋮ ⋮ ⋱ ⋮
𝟎 … … 𝝀𝒏
Let us define x=P z
𝒛 = 𝑷−𝟏 𝑨𝑷𝒛 + 𝑷−𝟏 𝑩𝑼

Take 𝑷−𝟏 𝑩 = 𝐅 = (𝒇𝒊𝒋 )


ALTERNATE FORM OF THE CONDITION FOR COMPLETE STATE
CONTROLLABILITY

The previous equation can be rewritten as follows:

𝒛𝟏=𝝀𝟏 𝒛𝟏 +𝒇𝟏𝟏𝒖𝟏 +𝒇𝟏𝟐 𝒖𝟐+⋯…+𝒇𝟏𝒓 𝒖𝒓


𝒛𝟐=𝝀𝟐 𝒛𝟐+𝒇𝟐𝟏𝒖𝟏 +𝒇𝟐𝟐 𝒖𝟐+⋯…+𝒇𝟐𝒓 𝒖𝒓

𝒛𝒏=𝝀𝒏 𝒛𝒏+𝒇𝒏𝟏𝒖𝟏 +𝒇𝒏𝟐 𝒖𝟐+⋯…+𝒇𝒏𝒓 𝒖𝒓
Then, if the elements of any row of F are all zeros, the system is
not completely state controllable.
ALTERNATE FORM OF THE CONDITION FOR COMPLETE STATE CONTROLLABILITY

If matrix A does not have distinct eigenvalues (vectors), then


diagonalization is impossible. In this case A can be transformed into
Jordan canonical form:
Example:-
A has eigenvalues 𝝀𝟏 , 𝝀𝟏 , 𝝀𝟏 , 𝝀𝟒, 𝝀𝟒 , 𝝀𝟔 , … … , 𝝀𝒏 and has n-3 distinct
eigenvalues, the Jordan canonical form is as follows:

𝝀𝟏 𝟏 𝟎 𝟎
𝟎 𝝀𝟏 𝟏
𝟎 𝟎 𝝀𝟏
𝝀𝟒 𝟏
𝑱=
𝟎 𝝀𝟒
𝝀𝟔

𝟎 𝝀𝒏
Let 𝑺−𝟏 𝑨𝑺=J, define x=S z
𝒛 = 𝑺−𝟏 𝑨𝑺 𝒛 + 𝑺−𝟏 𝑩 𝑼
= 𝑱 𝒛 + 𝑺−𝟏 𝑩 𝑼
Conditions of complete state controllability of the above system:
• No two Jordan blocks J are associated with the same eigenvalue
• The elements of any row of 𝑺−𝟏𝑩 that correspond to the last row of
each Jordan block are not always zero.
• The elements of each row of 𝑺−𝟏𝑩 that correspond to distinct
Eigenvalues are not always zero.
EXAMPLE I:- COMPLETELY STATE CONTROLLABLE SYSTEMS

𝒙𝟏 −𝟏 𝟎 𝒙𝟏 𝟐
= 𝒙 + 𝒖
𝒙𝟐 𝟎 −𝟐 𝟐 𝟓
𝒙𝟏 −𝟏 𝟏 𝟎 𝒙𝟏 𝟎
𝒙𝟐 = 𝟎 −𝟏 𝟎 𝒙𝟐 + 𝟒 𝒖
𝒙𝟑 𝟎 𝟎 −𝟐 𝒙𝟑 𝟑

𝒙𝟏 −𝟐 𝟏 𝟎 𝟎 𝒙𝟏 𝟎 𝟏
𝒙𝟐 𝟎 −𝟐 𝟏 𝒙𝟐 𝟎 𝟎 𝒖
𝒙𝟑 + 𝟑 𝟎 𝟏
𝒙𝟑 = 𝟎 𝟎 −𝟐 𝒖𝟐
𝒙𝟒 −𝟓 𝟏 𝒙𝟒 𝟎 𝟎
𝒙𝟓 𝟎 𝟎 −𝟓 𝒙𝟓 𝟐 𝟏
EXAMPLE II:- NOT COMPLETELY STATE CONTROLLABLE SYSTEMS

𝒙𝟏 −𝟏 𝟎 𝒙𝟏 𝟐
= + 𝒖
𝒙𝟐 𝟎 −𝟐 𝒙𝟐 𝟎

𝒙𝟏 −𝟏 𝟏 𝟎 𝒙𝟏 𝟒 𝟐 𝒖
𝟏
𝒙𝟐 = 𝟎 −𝟏 𝟎 𝒙𝟐 + 𝟎 𝟎 𝒖
𝟐
𝒙𝟑 𝟎 𝟎 −𝟐 𝒙𝟑 𝟑 𝟎

𝒙𝟏 −𝟐 𝟏 𝟎 𝟎 𝒙𝟏 𝟒
𝒙𝟐 𝟎 −𝟐 𝟏 𝒙𝟐 𝟐
𝒙𝟑 = 𝟎 𝟎 −𝟐 𝒙𝟑 + 𝟏 𝒖
𝒙𝟒 −𝟓 𝟏 𝒙𝟒 𝟑
𝒙𝟓 𝟎 𝟎 −𝟓 𝒙𝟓 𝟎
CONDITIONS FOR COMPLETE STATE CONTROLLABILITY IN THE S-PLANE
A necessary and sufficient condition for complete state controllability is that
no cancellation occurs in the transfer function or transfer matrix. If
cancellation occurs, the system can not be controlled in the direction of the
cancelled mode.
Example:-
𝑿(𝑺) (𝑺 + 𝟐. 𝟓)
=
𝑼(𝑺) (𝑺 + 𝟐. 𝟓)(𝑺 − 𝟏)
Because of the cancellation of the (S+2.5) term in the transfer matrix. This
system is not completely state controllable.
IN STATE SPACE

𝒙𝟏 𝟎 𝟏 𝒙𝟏 𝟏
= 𝒙 + 𝒖
𝒙𝟐 𝟐. 𝟓 −𝟏. 𝟓 𝟐 𝟏
Since:
𝟏 𝟏
𝑩 𝑨𝑩 =
𝟏 𝟏
The rank of 𝑩 𝑨𝑩 is 1. The system is not completely state controllable.
OUTPUT CONTROLLABILITY

𝑥 𝑡 = 𝑨 𝑥 𝑡 + 𝐵 U(t)
𝑦 = 𝑪 𝑥 𝑡 + 𝐷 U(t)
The system is said to be output controllable if it is possible to construct an
unconstrained control vector u(t) that will transfer any given initial output
𝒚 𝒕𝒐 to any final output 𝒚 𝒕𝒇 in a finite time interval 𝒕𝟎 ≤ 𝒕 ≤ 𝒕𝒇
The system is completely output controllable if and only if the mx(n+1)r
matrix.
𝑪𝑩: 𝑪𝑨𝑩: 𝑪𝑨𝟐 𝑩 … . . 𝑪𝑨𝒏−𝟏 𝑩: 𝑫 is of Rank m.
OBSERVABILITY

Consider the unforced system


𝑥 𝑡 =𝑨𝑥 𝑡
𝑦 =𝑪𝑥 𝑡
The system is said to be completely observable if every state 𝒙 𝒕𝒐 can be
determined from the observation of system 𝒚 𝒕𝒐 over a finite time interval
𝒕𝟎 ≤ 𝒕 ≤ 𝒕𝒇 . The system is then completely observable if every transition of
the state eventually affects every element of the output vector.
The output vector y(t) is:-
𝒚 𝒕 = 𝑪 𝒆𝑨𝒕 𝒙 𝒐
noting that
𝒏−𝟏
𝒆𝑨𝒕 = 𝒌=𝟎 ∝𝒌 𝒕 𝑨𝒌
𝒏−𝟏
𝒚 𝒕 = 𝑪 𝒆𝑨𝒕 𝒙 𝒐 = ∝𝒌 𝒕 𝑪 𝑨𝒌 𝒙 𝒐
𝒌=𝟎
OR
𝒚 𝒕 =∝𝒐 𝒕 𝑪 𝑿 𝟎 + ∝𝟏 𝒕 𝑪 𝑨 𝑿 𝒐 + ⋯ … . . +∝𝒏−𝟏 𝒕 𝑪 𝑨𝒏−𝟏 𝑿 𝒐

𝜶𝒐
𝜶
𝒚 𝒕 = 𝑪: 𝑪𝑨: … . . 𝑪𝑨𝒏−𝟏 … 𝟏. . 𝑿 𝒐
………
𝜶𝒏−𝟏
If the system is completely observable then, given the output y(t) over a time
interval 𝟎 ≤ 𝒕 ≤ 𝒕𝒇 , x(o) is uniquely determined from the above equation
It can be shown that this requires, the rank of the nmxn matrix

𝑪
𝑪𝑨

𝑪𝑨𝒏−𝟏
To be n or the nxnm matrix
𝑪∗ 𝑨∗ 𝑪∗ … (𝑨∗ )𝒏−𝟏 𝑪∗
Is of rank n or has n linearly independent column vectors.
This matrix is called the observability matrix
Consider the system described by:-
𝒙𝟏 𝟏 𝟏 𝒙𝟏 𝟎
= 𝒙 + 𝒖
𝒙𝟐 −𝟐 −𝟏 𝟐 𝟏
𝒙𝟏
𝒚= 𝟏 𝟎 𝒙
𝟐

Is it controllable and Observable?!


𝟎 𝟏
𝑩 … 𝑨𝑩 =
𝟏 −𝟏
Rank=2
Or the system is completely state controllable
𝑪𝑩 𝑪𝑨𝑩 = 𝟎 𝟏
Rank=1
Or the system is completely Output Controllable.
𝟏 𝟏
𝑪∗ 𝑨 ∗ 𝑪∗ =
𝟎 𝟏
The Rank of 𝑪∗ 𝑨∗ 𝑪∗ is 2

The System is completely observable.


CONDITIONS FOR COMPLETE OBSERVABILITY IN THE S-PLANE.

The necessary and sufficient condition for complete


observability is that no cancellation occurs in the
transfer function or transfer matrix. If cancellation
occurs in the transfer function, the canceled mode can
not be observed at the output.
RELATIONSHIPS BETWEEN CONTROLLABILITY, OBSERVABILITY, AND
TRANSFER FUNCTION

The transfer function has no cancelation if and only if


the system is completely state controllable and
completely observable.
Cancelled transfer function does not carry along all
information characterizing the dynamic system.
ALTERNATE FORM OF THE CONDITIONS FOR COMPLETE OBSERVABILITY

Let:
𝑥 𝑡 =𝑨𝑥 𝑡
𝑦 =𝑪𝑥 𝑡
Suppose:
𝑃−1 𝐴 𝑃 = 𝑫
Where D is a diagonal matrix
ALTERNATE FORM OF THE CONDITIONS FOR COMPLETE OBSERVABILITY CONT’D

Let
X=Pz
𝒛 = 𝑷−𝟏 𝑨𝑷 = 𝑫 𝒛
𝒚=𝒄𝑷𝒛
𝒚 𝒕 = 𝑪 𝑷 𝒆𝑫𝒕 𝒛 𝟎
Or
𝒆 𝝀𝟏 𝒕 𝟎 𝒆𝝀𝟏 𝒕 𝒛𝟏(𝟎)
𝒚 𝒕 =𝑪𝑷 𝒆𝝀𝟐 𝒕 𝒛 𝟎 =𝐂𝐏 𝒆 𝝀𝟐 𝒕
𝒛𝟐(𝟎)
⋱ …
𝟎 𝒆 𝝀𝒏 𝒕 𝒆𝝀𝒏 𝒕 𝒛𝒏(𝟎)
ALTERNATE FORM OF THE CONDITIONS FOR COMPLETE OBSERVABILITY CONT’D

• The system is completely observable if non of the columns of the mxn


matrix C P consists of all zero elements.
• If the ith column of C P consists of all zero elements then, the state
variable 𝒛𝒊 𝟎 will not appear in the output equation and therefore
can not be determined.
• This test applies only if 𝑷−𝟏AP is in diagonal form.
• If A has repeated eigenvalues, then the transformation matrix S
can be used to transfer A into Jordan Canonical form.
Let
𝒙=𝑺𝒛
𝒛 = 𝑺−𝟏 𝑨 𝑺 𝒛 = 𝑱𝒛
𝒚=𝑪𝑺𝒛
𝑱𝒕
𝒚 𝒕 = 𝑪 𝑺 𝒆 𝒛(𝟎)
THEN THE SYSTEM IS COMPLETELY OBSERVABLE IF

• No two Jordan blocks in J are associated with the same


eigenvalues.

• No columns of C S that correspond to the first row of each


Jordan block consists of zero elements.

• No columns of C S that correspond to distinct eigenvalues


consist of zero elements.
EXAMPLES:- COMPLETELY OBSERVABLE SYSTEMS

𝒙𝟏 −𝟏 𝟎 𝒙𝟏 𝒙𝟏
= 𝒙 ,𝒚= 𝟏 𝟑 𝒙
𝒙𝟐 𝟎 −𝟐 𝟐 𝟐

𝒙𝟏 𝟐 𝟏 𝟎 𝒙𝟏 𝒚𝟏
𝒙𝟏
𝒙𝟐 = 𝟎 𝟐 𝒙 𝟑 𝟎 𝟎 𝒙
𝟐 ,
𝟏 𝒚𝟐 = 𝟒 𝟎 𝟎 𝒙
𝟐
𝒙𝟑 𝟎 𝟎 𝟐 𝒙𝟑 𝟑

𝒙𝟏 𝟐 𝟏 𝟎 𝟎 𝒙𝟏 𝒙𝟏
𝒙𝟐 𝟎 𝟐 𝟏 𝒙𝟐 𝒙𝟐
𝒚𝟏 𝟏 𝟏 𝟏 𝟎 𝟎 𝒙𝟑
𝒙𝟑 = 𝟎 𝟎 𝟐 𝒙𝟑 , =
𝒚𝟐 𝟎 𝟏 𝟏 𝟏 𝟎 𝒙𝟒
𝒙𝟒 −𝟑 𝟏 𝒙𝟒
𝒙𝟓 𝟎 𝟎 −𝟑 𝒙𝟓 𝒙𝟓
EXAMPLES:- SYSTEMS THAT ARE NOT COMPLETELY OBSERVABLE

𝒙𝟏 −𝟏 𝟎 𝒙𝟏 𝒙𝟏
= 𝒙 , 𝒚= 𝟎 𝟏 𝒙
𝒙𝟐 𝟎 −𝟐 𝟐 𝟐

𝒙𝟏 𝟐 𝟏 𝟎 𝒙𝟏 𝒚𝟏
𝒙𝟏
𝒙𝟐 = 𝟎 𝟐 𝒙 𝟎 𝟏 𝟑 𝒙
𝟐 ,
𝟏 𝒚𝟐 = 𝟎 𝟐 𝟒 𝒙
𝟐
𝒙𝟑 𝟎 𝟎 𝟐 𝒙𝟑 𝟑

𝒙𝟏 𝟐 𝟏 𝟎 𝟎 𝒙𝟏 𝒙𝟏
𝒙𝟐 𝟎 𝟐 𝟏 𝒙𝟐 𝒙𝟐
𝒚
𝒙𝟑 = 𝟎 𝟎 𝟐 𝒙𝟑 , 𝟏 = 𝟏 𝟏 𝟏 𝟎 𝟎 𝒙𝟑
𝒚𝟐 𝟎 𝟏 𝟏 𝟎 𝟎 𝒙𝟒
𝒙𝟒 −𝟑 𝟏 𝒙𝟒
𝒙𝟓 𝟎 𝟎 −𝟑 𝒙𝟓 𝒙𝟓

You might also like