Scribd
Upload
37 views

Lecture-19 - Stability of LTI Systems

The document discusses stability analysis of linear time-invariant systems. It begins by defining stability as a system returning to equilibrium after a disturbance. The nature of solutions is determined by the locations of the system's poles (eigenvalues of the A matrix). Stability depends on whether the poles lie in the left or right half planes. For systems with n linearly independent eigenvectors, solutions are given by the eigenvalues. In general cases, Jordan forms are used. Stabilizing controllers place all closed-loop poles in the left half plane for stability. Methods to compute the matrix exponential eAt include power series, Laplace transforms, similarity transforms, and Sylvester's formula.

Uploaded by

JS Channel
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
37 views

Lecture-19 - Stability of LTI Systems

The document discusses stability analysis of linear time-invariant systems. It begins by defining stability as a system returning to equilibrium after a disturbance. The nature of solutions is determined by the locations of the system's poles (eigenvalues of the A matrix). Stability depends on whether the poles lie in the left or right half planes. For systems with n linearly independent eigenvectors, solutions are given by the eigenvalues. In general cases, Jordan forms are used. Stabilizing controllers place all closed-loop poles in the left half plane for stability. Methods to compute the matrix exponential eAt include power series, Laplace transforms, similarity transforms, and Sylvester's formula.

Uploaded by

JS Channel
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/ 15

Lecture – 19

Stability of Linear Time Invariant Systems

Prof. Radhakant Padhi


Dept. of Aerospace Engineering
Indian Institute of Science - Bangalore

Stability of Linear Systems


Definition: If a system in equilibrium is disturbed
and the system returns back to the equilibrium
point with time, then the equilibrium point is said to
be stable.

Prof. Radhakant Padhi, IISc-Bangalore 2

1
Stability of Linear Time
Invariant (LTI) Systems
System:

Question:
Can we conclude about nature of the solution, without
solving the system model?
Answer: YES!

Definition: Eigenvalues of A : “Poles” of the system!

The nature of the solution is governed only by the


locations of its poles
Prof. Radhakant Padhi, IISc-Bangalore 3

Similarity Transformation
Definition:

Simplest forms possible:


• Diagonal form
(if there are n linearly independent eigenvectors)
• Jordan form
(if the number of linearly independent eigenvectors are
less then n)

Prof. Radhakant Padhi, IISc-Bangalore 4

2
Stability Analysis - Special Case:
Anxn has n linearly independent eigenvectors

Prof. Radhakant Padhi, IISc-Bangalore 5

Special Case:
Anxn has n linearly independent eigenvectors

Solution:

Prof. Radhakant Padhi, IISc-Bangalore 6

3
Special Case:
Anxn has n linearly independent eigenvectors

Solution:

Conclusion
The nature of solution depends only on the
location of poles!

All poles in the LH plane: Stable System


One pole in the RH plane: Unstable System
Prof. Radhakant Padhi, IISc-Bangalore 7

General Case: Anxn does not have n


linearly independent eigenvectors

Prof. Radhakant Padhi, IISc-Bangalore 8

4
General Case: Anxn does not have n
linearly independent eigenvectors

Solution:

Prof. Radhakant Padhi, IISc-Bangalore 9

General Case: Anxn does not have n


linearly independent eigenvectors

Prof. Radhakant Padhi, IISc-Bangalore 10

5
General Case: Anxn does not have n
linearly independent eigenvectors

Prof. Radhakant Padhi, IISc-Bangalore 11

General Case: Anxn does not have n


linearly independent eigenvectors

Prof. Radhakant Padhi, IISc-Bangalore 12

6
General Case: Anxn does not have n
linearly independent eigenvectors

Exponential term will eventually dominate the polynomial term!

Prof. Radhakant Padhi, IISc-Bangalore 13

Stability of Linear Systems

Conclusion

The nature of solution depends only on the


location of poles!

All poles in the LH plane: Stable System


One pole in the RH plane: Unstable System

Prof. Radhakant Padhi, IISc-Bangalore 14

7
Stabilizing Control Design
Closed loop system:

 Closed loop system is stable if Eigenvalues


of satisfy the stability condition

 For stabilizing controller needs to be


selected in such a way that the eigenvalues
of should be in the left half plane
Prof. Radhakant Padhi, IISc-Bangalore 15

Evaluation of Matrix Exponential eAt

Prof. Radhakant Padhi


Dept. of Aerospace Engineering
Indian Institute of Science - Bangalore

8
Method – 1: Power-series

 This method is useful and accurate only if the


series truncates naturally. Otherwise, series
truncation introduces approximation error.
 Direct computation of eAt as power series is
computationally inefficient as well.

Prof. Radhakant Padhi, IISc-Bangalore 17

Method – 2:
Using Laplace Transform
Problem:
Solution using Laplace transform:

Solution known: Comparing the two solutions:

Prof. Radhakant Padhi, IISc-Bangalore 18

9
Evaluation of :
How to compute it efficiently?

Prof. Radhakant Padhi, IISc-Bangalore 19

Symbolic Computation of

Equate the coefficients on both sides…


Prof. Radhakant Padhi, IISc-Bangalore 20

10
Symbolic Computation of

This suggests a recursive algorithm!


(provided are known)

Reference: T. Kailath, Linear Systems, Prentice-Hall, 1980.

Prof. Radhakant Padhi, IISc-Bangalore 21

Method – 2:
Using Laplace Transform

 This method results in closed form expressions for


eAt, can be quite useful for small matrices.
 Numerical algorithms exist to evaluate .
However, Laplace inverse of each element still
need to be found.
 Can be quite cumbersome for large matrices.

Prof. Radhakant Padhi, IISc-Bangalore 22

11
Method – 3:
Using Similarity Transform
(Provided the matrix can be diagonalizable)

Similarity
Transformation:

Prof. Radhakant Padhi, IISc-Bangalore 23

Method – 4: Sylvester’s Formula


Case – 1: Distinct Eigenvalues

eAt satisfies the following determinant equation:

Expanding by the last row,

Prof. Radhakant Padhi, IISc-Bangalore 24

12
Method – 4: Sylvester’s Formula
Case – 1: Distinct Eigenvalues

Prof. Radhakant Padhi, IISc-Bangalore 25

Method – 4: Sylvester’s Formula


Case – 2: Repeated Eigenvalues

eAt satisfies the following determinant equation:

i.e.

Prof. Radhakant Padhi, IISc-Bangalore 26

13
Method – 4: Sylvester’s Formula
Case – 2: Repeated Eigenvalues

Prof. Radhakant Padhi, IISc-Bangalore 27

Example: Sylvester’s Formula

Prof. Radhakant Padhi, IISc-Bangalore 28

14
Prof. Radhakant Padhi, IISc-Bangalore 29

15

You might also like