Stability and Phase Plane

Analysis

1
Advanced Control (Mehdi Keshmiri, Winter 95)
 Objectives

Objectives of the section:

 Introducing the Phase Plane Analysis

 Introducing the Concept of stability

 Stability Analysis of Linear Time Invariant Systems

 Lyapunov Indirect Method in Stability Analysis of

Nonlinear Systems

2
Advanced Control (Mehdi Keshmiri, Winter 95)
 Introducing
the Phase Plane Analysis

3
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Phase Space form of a Dynamical System:

𝑥1 = 𝑓1 𝑥1 , 𝑥2 , … , 𝑥𝑛 , 𝑢1 , 𝑢2 , … , 𝑢𝑚 , 𝑡
𝑥2 = 𝑓2 (𝑥1 , 𝑥2 , … , 𝑥𝑛 , 𝑢1 , 𝑢2 , … , 𝑢𝑚 , 𝑡) 𝑋 = 𝐹 𝑋, 𝑈, 𝑡
⋮ 𝑋 ∈ ℝ𝑛 𝑈 ∈ ℝ𝑚
𝑥𝑛 = 𝑓𝑛 𝑥1 , 𝑥2 , … , 𝑥𝑛 , 𝑢1 , 𝑢2 , … , 𝑢𝑚 , 𝑡

𝑼 𝑿 𝑼 𝑿
𝑋 = 𝐹 𝑋, 𝑈, 𝑡 𝑋 = 𝐹 𝑋, 𝑈

Time-Varying System Time-Invariant System

4
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Phase Space form of a Linear Time Invariant (LTI) System:

𝑿 = 𝑨𝑿 + 𝑩𝑼
𝑋 ∈ ℝ𝑛 𝑈 ∈ ℝ𝑚

Special Properties of Nonlinear Systems:

 Multiple isolated equilibria

 Limit Cycle
 Finite escape time
 Harmonic, sub-harmonic and almost periodic Oscillation
 Chaos
 Multiple modes of behavior

5
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Phase Plane Analysis is a graphical method for studying

second-order systems respect to initial conditions by:
 providing motion trajectories corresponding to various initial
conditions.

 examining the qualitative features of the trajectories

 obtaining information regarding the stability of the equilibrium

points

𝑥1 = 𝑓1 (𝑥1 , 𝑥2 )

𝑥2 = 𝑓2 (𝑥1 , 𝑥2 )

6
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Advantages of Phase Plane Analysis:

 It is graphical analysis and the solution trajectories can be

represented by curves in a plane

 Provides easy visualization of the system qualitative

 Without solving the nonlinear equations analytically, one can study

the behavior of the nonlinear system from various initial
conditions.

 It is not restricted to small or smooth nonlinearities and applies

equally well to strong and hard nonlinearities.

 There are lots of practical systems which can be approximated by

second-order systems, and apply phase plane analysis.

7
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Disadvantage of Phase Plane Method:

 It is restricted to at most second-order

 graphical study of higher-order is computationally and

geometrically complex.

8
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Example: First Order LTI System

𝒙 = 𝐬𝐢𝐧(𝒙)

Analytical Solution Graphical Solution

𝑑𝑥 sin(x)
= sin(𝑥) 1
𝑑𝑡
0.5
𝑑𝑥 0
= 𝑑𝑡
sin(𝑥) -0.5

-1
𝑥 𝑡
𝑑𝑥 -6 -4 -2 0 2 4 6
= 𝑑𝑡
𝑥0 sin(𝑥) 0
x

cos 𝑥0 + cot(𝑥0 )
𝑡 = ln
cos 𝑥 + cot(𝑥)

9
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis
Concept of Phase Plane Analysis:
 Phase plane method is applied to Autonomous Second Order System

𝑥1 = 𝑓1 (𝑥1 , 𝑥2 ) 𝑥2 = 𝑓2 (𝑥1 , 𝑥2 )

 System response 𝑋 𝑡 = (𝑥1 𝑡 , 𝑥2 (𝑡)) to initial condition 𝑋0 = 𝑥1 0 , 𝑥2 0

is a mapping from ℝ(Time) to ℝ2 (𝑥1 , 𝑥2 )
 The solution can be plotted in the 𝑥1 − 𝑥2 plane called State Plane or
Phase Plane
 The locus in the 𝑥1 − 𝑥2 plane is a curved named Trajectory that pass through
point 𝑋0
 The family of the phase plane trajectories corresponding to various initial
conditions is called Phase portrait of the system.
 For a single DOF mechanical system, the phase plane is in fact is (𝑥, 𝑥) plane
10
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Example: Van der Pol Oscillator Phase Portrait

x


11
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Plotting Phase Plane Diagram:

Analytical Method
Numerical Solution Method
Isocline Method
Vector Field Diagram Method
Delta Method
Lienard’s Method
Pell’s Method

12
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Analytical Method
 Dynamic equations of the system is solved, then time parameter is omitted
to obtain relation between two states for various initial conditions

𝑥1 = 𝑓1 (𝑥1 , 𝑥2 )

𝑥2 = 𝑓2 (𝑥1 , 𝑥2 )

Solve 𝑥1 𝑡, 𝑋0 = 𝑔1 (𝑡, 𝑋0 )
𝐹 𝑥1 , 𝑥2 = 0
𝑥2 𝑡, 𝑋0 = 𝑔2 (𝑡, 𝑋0 )

 For linear or partially linear systems

13
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Example: Mass Spring System

𝑚𝑥 + 𝑘𝑥 = 0

For 𝑚 = 𝑘 = 1 : 𝑥+𝑥 =0 x2 + y2 - 4

𝑥 𝑡 = 𝑥0 cos(𝑡) + 𝑥0 sin(𝑡) 1

𝑥 𝑡 = −𝑥0 sin 𝑡 + 𝑥0 cos(𝑡)

0

y
-1

-2
2
𝑥 +𝑥 =2
𝑥02 + 𝑥02 -2 -1 0 1 2
x

14
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Analytical Method
 Time differential is omitted from dynamic equations of the system, then
partial differential equation is solved

𝑥1 = 𝑓1 (𝑥1 , 𝑥2 )

𝑥2 = 𝑓2 (𝑥1 , 𝑥2 )

𝑑𝑥2 𝑓2 (𝑥1 , 𝑥2 ) Solve

= 𝐹 𝑥1 , 𝑥2 = 0
𝑑𝑥1 𝑓1 (𝑥1 , 𝑥2 )

 For linear or partially linear systems

15
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis
Example: Mass Spring System

𝑚𝑥 + 𝑘𝑥 = 0

𝑥1 = 𝑥2
For 𝑚 = 𝑘 = 1 : 𝑥+𝑥 =0 𝑥2 = −𝑥1
𝑑𝑥2 −𝑥1 x2 + y2 - 4
=
𝑑𝑥1 𝑥2 2

𝑥2 𝑥1 1

𝑥2 𝑑𝑥2 = −𝑥1 𝑑𝑥1 0

y
𝑥20 𝑥10
-1

𝑥 2 + 𝑥 2 = 𝑥02 + 𝑥02
-2
-2 -1 0 1 2
x
16
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Numerical Solution Method

Dynamic equations of the system is solved numerically (e.g. Ode45) for
various initial conditions and time response is obtained, then two states are
plotted in each time.

Example: Pendulum 𝜃 + sin 𝜃 = 0

1
1 1

0.5
0.5 0.5
x 1(t)

x 2(t)

x2
0 0

-0.5 -0.5 -0.5

-1 -1
0 2 4 6 8 10 0 2 4 6 8 10 -1
Time(sec) Time(sec) -1 -0.5 0 0.5 1
x1

17
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Isocline Method
Isocline: The set of all points which have same trajectory slope

𝑥1 = 𝑓1 (𝑥1 , 𝑥2 )

𝑥2 = 𝑓2 (𝑥1 , 𝑥2 )

𝑑𝑥2 𝑓2 (𝑥1 , 𝑥2 )
= =𝛼 𝑓2 𝑥1 , 𝑥2 = 𝛼𝑓1 (𝑥1 , 𝑥2 )
𝑑𝑥1 𝑓1 (𝑥1 , 𝑥2 )

First various isoclines are plotted, then trajectories are drawn.

18
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis
Example: Mass Spring System

𝑚𝑥 + 𝑘𝑥 = 0

𝑥1 = 𝑥2
For 𝑚 = 𝑘 = 1 : 𝑥+𝑥 =0 𝑥22 =2 −𝑥1
x +y -4
5
Slope=1
𝑑𝑥2 −𝑥1
= =𝛼 Slope=-1
𝑑𝑥1 𝑥2
0
y

Slope=infinite

𝑥1 + 𝛼𝑥2 = 0

-5
-5 0 5
x
19
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Vector Field Diagram Method

Vector Field: A set of vectors that is tangent to the trajectory

𝑓1 (𝑥1 , 𝑥2 )
 At each point (𝑥1 , 𝑥2 ) vector is tangent to the trajectories
𝑓2 (𝑥1 , 𝑥2 )

 Hence vector field can be constructed in the phase plane and direction of
the trajectories can be easily realized with that

  sin( )  0

x1     x2 
f   
x2      sin( x1 
)

20
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis
Singular Points in the Phase Plane Diagram:
Equilibrium points are in fact singular points in the phase plane
diagram
𝑓1 𝑥1 , 𝑥2 = 0 Slope of the trajectories at 𝑑𝑥2 0
=
𝑓1 𝑥1 , 𝑥2 = 0
equilibrium points 𝑑𝑥1 0

 Singular point is an important concept which reveals great info about

properties of system such as stability.

 Singular points are only points which several trajectories pass/approach

them (i.e. trajectories intersect).

21
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Example: Using Matlab

𝑥 + 0.6𝑥 + 3𝑥 + 𝑥 2 = 0
x '=y
y ' = - 0.6 y - 3 x + x 2
10

y
-5

-10

-6 -4 -2 0 2 4 6
x

22
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Example: Using Maple Code

with(DEtools):
𝑥 + 0.6𝑥 + 3𝑥 + 𝑥 2 = 0
xx:=x(t): yy:=y(t):
dx:=diff(xx,t): dy:=diff(yy,t):
e0:=diff(dx,t)+.6*dx+3*xx+xx^2:
e1:=dx-yy=0:
e2:=dy+0.6*yy+3*xx+xx^2=0:
eqn:=[e1,e2]: depvar:=[x,y]:
rang:=t=-1..5: stpsz:=stepsize=0.005:
IC1:=[x(0)=0,y(0)=1]:
IC2:=[x(0)=0,y(0)=5]:
IC3:=[x(0)=0,y(0)=7]:
IC4:=[x(0)=0,y(0)=7]:
IC5:=[x(0)=-3.01,y(0)=0]:
IC6:=[x(0)=-4,y(0)=2]:
IC7:=[x(0)=1,y(0)=0]:
IC8:=[x(0)=4,y(0)=0]:
IC9:=[x(0)=-6,y(0)=3]:
IC10:=[x(0)=-6,y(0)=6]:
ICs:=[IC||(1..10)]:
lincl:=linecolour=sin((1/2)*t*Pi):
mtd:=method=classical[foreuler]:
phaseportrait(eqn,depvar,rang,ICs,stpsz,lincl,mtd);

23
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Example: Using Maple Tools

𝑥 + 0.6𝑥 + 3𝑥 + 𝑥 2 = 0

24
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis

Phase Plane Analysis for Single DOF Mechanical System

In the case of single DOF mechanical system

𝑥1 = 𝑥 𝑥1 = 𝑥2
𝑥 + 𝑔 𝑥, 𝑥 = 0 𝑥2 = 𝑥 𝑥2 = −𝑔(𝑥1 , 𝑥2 )

 The phase plane is in fact (𝑥 − 𝑥) plane and every point shows the
position and velocity of the system.

 Trajectories are always clockwise. This is not true in the general phase
plane (𝑥1 − 𝑥2 )

25
Advanced Control (Mehdi Keshmiri, Winter 95)
 Introducing
the Concept of Stability

26
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples

 Stability analysis of a dynamic system is normally introduced in the state

space form of the equations.
𝑋 = 𝐹 𝑋, 𝑈, 𝑡
𝑋 ∈ ℝ𝑛 𝑈 ∈ ℝ𝑚

Dynamic System

u x = f(x,u,t ) x

 Most of the concepts in this chapter are introduced for autonomous

systems
Autonomous Dynamic System

u x = f(x,u) x

27
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples
Stability analysis of a dynamic system is divided in three
categories:
1. Stability analysis of the equilibrium points of the systems. We study the
behavior (dynamics) of the free (unforced, 𝑢 = 0) system when it is perturbed
from its equilibrium point.

2. Input-output stability analysis. We study the system (forced system 𝑢 ≠ 0)

output behavior in response to bounded inputs.

3. Stability analysis of periodic orbits. This analysis is for those systems which
perform a periodic or cyclic motion like walking of a biped or orbital motion
of a space object.

 Our main concern is the first type analysis. Some preliminary issues of the
second type analysis will be also discussed.

28
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples

Reminder:
𝑋𝑒 is said an equilibrium point of the system if once the system reaches this
position it stays there for ever, i.e. 𝑓 𝑋𝑒 = 0

Definition (Lyapunov Stability):

The equilibrium point 𝑋𝑒 is said to be stable (in the sense of Lyapunov
stability) or motion of the system about its equilibrium point is said to be
stable if the system states (𝑋) is perturbed away from 𝑋𝑒 then it stays close to
𝑋𝑒 . Mathematically 𝑋𝑒 is stable if

  0 ,     ( )  0 x (0)  xe    x(t )  xe   t  0

29
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples
 Without loss of generality we can present our analysis about equilibrium
point 𝑋𝑒 = 0, since the system equation can be transferred to a new form
with zero as the equilibrium point of the system.

𝑌=𝐹 𝑌 𝑋 = 𝑌 − 𝑌𝑒 𝑋=𝐹 𝑋
𝑌𝑒 ≠ 0 𝑋𝑒 ≠ 0

A more precise definition:

The equilibrium point 𝑋𝑒 is said to be stable (in the sense of Lyapunov
stability) or motion of the system about its equilibrium point is said to be stable
if for any 𝑅 > 0, there exists 0 < 𝑟 < 𝑅 such that

x (0)  xe  r  x(t )  xe  R t  0

30
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples
Definition (Lyapunov Stability):
The equilibrium point 𝑋𝑒 = 0 is said to be
Stable if
∀𝑅 > 0 ∃0 < 𝑟 < 𝑅 𝑠. 𝑡.
𝑋 0 <𝑟 𝑋 𝑡 < 𝑅 ∀𝑡 > 0
Unstable if it is not stable.
Asymptotically stable if it is stable and
∀𝑟 > 0 𝑠. 𝑡.
𝑋 0 <𝑟 lim 𝑋(𝑡) = 0
𝑡→∞

Marginally stable if it is stable and not

asymptotically stable

Exponentially stable if it is asymptotically stable with an exponential rate

𝑋(0) < 𝑟 𝑋(𝑡) < 𝛼𝑒 −𝛽𝑡 𝑋(0) 𝛼, 𝛽 > 0
31
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples

Example: Undamped Pendulum

𝑔
𝜃 + sin(𝜃) = 0 𝜃𝑒1 = 0 , 𝜃𝑒2 = 𝜋
𝑙
 𝜃𝑒1 is a marginally stable point and 𝜃𝑒2 is an unstable point
1.5

1 X(0) = (0,1)

0.5

Teta (rad)
0

-0.5

Br BR -1

-1.5
0 2 4 6 8 10
   Time(sec)

40

30
X(0) = (0,4)

Teta (rad)
20

10

0
0 2 4 6 8 10
Time(sec)
32
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples

Example: damped Pendulum

𝑔
𝜃 + 𝐶 𝜃 + sin(𝜃) = 0 𝜃𝑒1 = 0 , 𝜃𝑒2 = 𝜋
𝑙
 𝜃𝑒1 is an exponentially stable point and 𝜃𝑒2 is an unstable point
0.5

-0.5

Teta (rad)
-1 X(0) = (-2.5,0)

-1.5

-2 -3

-2.5
0 5 10 15 -4 X(0) = (-3.5,0)
Time(sec)

Teta (rad)
-5

-6
8

6
-7
0 5 10 15
4
Teta (rad)

Time (sec)
X(0) = (-3,10)
2

-2

-4
0 5 10 15
Time(sec)
33
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples

Example: Van Der Pol Oscillator

stateform
 x1  x2
x  (1  x) x  x  0   x1e  x2e  0
2

 x2   x1  (1  x1 ) x2
2

 𝑥𝑒 = 0 is an unstable point

34
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples

Definition:

if the equil. point 𝑋𝑒 is asymptotically stable, then the set of all

points that trajectories initiated at these point eventually converge
to the origin is called domain of attraction.

Definition:

if the equil. point 𝑋𝑒 is asymptotically/exponentially stable, then

the equil. point is called globally stable if the whole space is
domain of attraction. Otherwise it is called locally stable.

35
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples

Example 1:
The origin in the first order system of 𝑥 = −𝑥 is globally exponentially stable.

x   x  x(t )  x0 et  lim x(t )  0  x0  0

t 

Example 2:
The origin in the first order system 𝑥 = −𝑥 3 is globally asymptotically but
not exponentially stable.

x0
x   x  x(t ) 
3
 lim x(t )  0  x0
t 
1  2tx 2
0

36
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples

Example 3:

The origin in the first order system 𝑥 = −𝑥 2 is semi-asymptotically but not

exponentially stable.

lim x(t )  0 if x0  0
x t 
x   x 2  x(t )  0  
1  tx0 lim x(t )   if x0  0
t 1/ x 0

Domain of attraction is 𝑥0 > 0.

37
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples

Example 4:
𝑥+𝑥+𝑥 =0

38
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples

Example 5:

𝑥 + 0.6𝑥 + 3𝑥 + 𝑥 2 = 0

39
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability, Definitions and Examples

Example 6:
𝑥 + 𝑥 + 𝑥3 − 𝑥 = 0

40
Advanced Control (Mehdi Keshmiri, Winter 95)
 Stability Analysis of

Linear Time Invariant Systems

41
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

It is the best tool for study of the linear system graphically

This analysis gives a very good insight of linear systems

behavior

The analysis can be extended for higher order linear system

Local behavior of the nonlinear systems can be understood from

this analysis

The analysis is performed based on the system eigenvalues and

eigenvectors.

42
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Consider a second order linear system:

𝑥 = 𝐴𝑥 𝐴 ∈ ℝ2×2 , 𝑥 ∈ ℝ2

 If the A matrix is nonsingular, origin is the only equilibrium

point of the system
A is non-singular  xe  0

 If the A matrix is singular then the system has infinite number

of equilibrium points. In fact all of the points belonging to the
null space of A are the equilibrium point of the system.

A issingular  xe  x* | x*  Null (A)

43
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems
Consider a second order linear system:

𝑥 = 𝐴𝑥
𝐴 ∈ ℝ2×2 , 𝑥 ∈ ℝ2

The analytical solution can be obtained based on eigenvalues (𝜆1 , 𝜆2 ):

If 𝜆1 , 𝜆2 are real and distinct 𝑥 𝑡 = 𝐴𝑒 𝜆1 𝑡 + 𝐵𝑒 𝜆2 𝑡

If 𝜆1 , 𝜆2 are real and similar 𝑥 𝑡 = (𝐴 + 𝐵𝑡)𝑒 𝜆𝑡

If 𝜆1 , 𝜆2 are complex conjugate 𝑥 𝑡 = 𝐴𝑒 𝜆1 𝑡 + 𝐵𝑒 𝜆2 𝑡

= 𝑒 𝛼𝑡 (𝐴 sin 𝛽𝑡 + 𝐵 cos(𝛽𝑡))

44
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Jordan Form (almost diagonal form)

This representation has the system eigenvalues on the leading diagonal, and
either 0 or 1 on the super diagonal.

𝑥 = 𝐴𝑥 𝑦 = 𝐽𝑦

𝐽1 𝜆𝑖 1
𝐽= ⋱ 𝐽𝑖 = ⋱ 1
𝐽𝑛 𝜆𝑖

Obtaining Jordan form:

𝑦 = 𝑃−1 𝑥
𝐽 = 𝑃−1 𝐴𝑃
𝑃 = 𝑣1 … 𝑣𝑛

45
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

A is non-singular:

The A matrix has 2 eigenvalues (either two real, or two complex

conjugates) and can have either two eigenvectors or one
eigenvectors. Four categories can be realized

1. Two distinct real eigenvalues and two real eigenvectors

2. Two complex conjugate eigenvalues and two complex

eigenvectors

3. Two similar (real) eigenvalues and two eigenvectors

4. Two similar (real) eigenvalues and one eigenvectors

46
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

1. Two distinct real eigenvalues and two real eigenvectors

𝜆1 0
𝑦 = 𝐽𝑦 𝐽=
0 𝜆2
𝑦1 = 𝜆1 𝑦1 𝑦1 = 𝑦10 𝑒 𝜆1𝑡

𝑦2 = 𝜆2 𝑦2 𝑦2 = 𝑦20 𝑒 𝜆2𝑡

𝑦1
ln = 𝜆1 𝑡 𝜆2
𝑦10
𝜆2 𝑦1 𝑦2 𝑦1 𝜆1 𝑦2
ln = ln =
𝑦2 𝜆1 𝑦10 𝑦20 𝑦10 𝑦20
ln = 𝜆2 𝑡
𝑦20

𝑦20
𝑦2 = 𝜆2 /𝜆1
𝑦1 𝜆2 /𝜆1 𝑦2 = 𝐾𝑦1 𝜆2/𝜆1
𝑦10

47
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

1. A ) 𝝀𝟐 < 𝝀𝟏 < 𝟎 𝑦2 = 𝐾𝑦1 𝜆2/𝜆1

 System has two eigenvectors 𝑣1 , 𝑣2 the phase plane portrait is as the

following
x'=-2x x'=y
y ' = - 12 y y'=-2x-3y

10 4

8
3
6
2
4

2
1

0 0

y
y

-2
-1
-4
-2
-6
-3
-8

-10 -4

-10 -8 -6 -4 -2 0 2 4 6 8 10 -4 -3 -2 -1 0 1 2 3 4
x x

 Trajectories are:
 tangent to the slow eigenvector (𝑣1 ) for near the origin
 parallel to the fast eigenvector (𝑣2 ) for far from the origin
 The equilibrium point 𝑋𝑒 = 0 is called stable node
48
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems
1.5
X(0) = [-1,10]

Example 7: 𝑥 + 4𝑥 + 2𝑥 = 0 1 X(0) = [-1,1]

0.5

0 1 0

𝑋= 𝑋
−2 −4
-0.5

-1
0 2 4 6 8 10

0 1
det 𝜆𝐼 − = 𝜆2 + 4𝜆 + 2 = 0 𝜆1 = −0.59 , 𝜆2 = −3.41
−2 −4

−0.59 −1 𝑣11 4
2 −0.59 + 4 2 =0
𝑣1
2
0.86
𝑣1 = 0
−0.51
y

−3.41 −1 𝑣11 -2

2 −3.41 + 4 2 =0
𝑣1 -4
-4 -2 0 2 4
−0.28
𝑣2 = x
0.96
49
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

1. B ) 𝝀𝟐 > 𝝀𝟏 > 𝟎 𝑦2 = 𝐾𝑦1 𝜆2/𝜆1

 System has two eigenvectors 𝑣1 and 𝑣2 the phase plane portrait is

opposite as the previous one x'=-2x
y ' = - 12 y
x'=y
y'=-2x-3y

10 4

8
3
6
2
4

2
1

0 0

y
y

-2
-1
-4
-2
-6
-3
-8

-10 -4

-10 -8 -6 -4 -2 0 2 4 6 8 10 -4 -3 -2 -1 0 1 2 3 4

 Trajectories are:
x x

 tangent to the slow eigenvector 𝑣1 for near origin

 parallel to the fast eigenvector 𝑣2 for far from origin

 The equilibrium point 𝑋𝑒 = 0 is called unstable node

50
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems
6

Example 8: 𝑥 − 3𝑥 + 2𝑥 = 0
X(0) = [-2,2]
4
X(0) = [-2,0.1]
2

0 1 0

𝑋= 𝑋
−2 3 -2

-4
0 0.2 0.4 0.6 0.8 1

0 1
det 𝜆𝐼 − = 𝜆2 − 3𝜆 + 2 = 0 𝜆1 = 1 , 𝜆2 = 2
−2 3

1 −1 𝑣11
2 =0
4
2 1−3 𝑣1
2
−0.71
𝑣1 =
−0.71 0
y

2 −1 𝑣11 -2
2 2−3 2 =0
𝑣1
-4
-4 -2 0 2 4
−0.45
𝑣2 = x
−0.89
51
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

1. C ) 𝝀𝟐 < 𝟎 < 𝝀𝟏 𝑦2 = 𝐾𝑦1 𝜆2/𝜆1

 System has two eigenvectors 𝑣1 and 𝑣2 , the phase plane portrait is as the
following x'=3x
y'=-3y
x'=y
y'=y+2x

4 4

3 3

2 2

1 1

0 0
y

y
-1 -1

-2 -2

-3 -3

-4 -4

-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
x x

 Only trajectories along 𝑣2 are stable trajectories

 All other trajectories at start are tangent to 𝑣2 and at the end are tangent
to 𝑣1

 This equilibrium point is unstable and is called saddle point

52
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems
8
x 10
3

Example 9: 𝑥 − 𝑥 − 2𝑥 = 0 2
X(0) = [1,0.5]
X(0) = [1,-1.5]

0 1
1

𝑋= 𝑋
2 1
0

-1
0 2 4 6 8 10

0 1
det 𝜆𝐼 − = 𝜆2 − 𝜆 − 2 = 0 𝜆1 = −1 , 𝜆2 = 2
2 1

−1 −1 𝑣11
2 =0
4
−2 −1 − 1 𝑣1
2
−0.71
𝑣1 =
0.71 0
y

2 −1 𝑣11 -2
−2 2 − 1 2 =0
𝑣1
-4
-4 -2 0 2 4
−0.45
𝑣2 = x
−0.89
53
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems
2. Two complex conjugate eigenvalues and two complex eigenvectors
𝛼 −𝛽
𝑦 = 𝐽𝑦 𝐽=
𝛽 𝛼

𝑟 ≡ 𝑦12 + 𝑦22

𝑦2
𝜃 ≡ tan−1 ( )
𝑦1

𝑟𝑟 = 𝑦1 𝑦1 + 𝑦2 𝑦2 = 𝑦1 𝛼𝑦1 − 𝛽𝑦2 + 𝑦2 𝛽𝑦1 + 𝛼𝑦2 = 𝛼𝑟 2

𝑦1 𝑦2 − 𝑦2 𝑦1 𝑦1 𝛽𝑦1 + 𝛼𝑦2 − 𝑦2 𝛼𝑦1 − 𝛽𝑦2

𝜃 1 + tan 𝜃 2 = 2 = 2 = 𝛽 1 + tan 𝜃 2
𝑦1 𝑦1

𝑟 = 𝛼𝑟
𝑟 𝑡 = 𝑟0 𝑒 𝛼𝑡 𝜃 𝑡 = 𝜃0 + 𝛽𝑡
𝜃=𝛽

54
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

2. A ) 𝝀𝟐 , 𝝀𝟏 = 𝜶 ± 𝜷𝒊 , 𝜶 < 𝟎 ,𝜷 ≠ 𝟎 𝑟 𝑡 = 𝑟0 𝑒 𝛼𝑡

𝜃 𝑡 = 𝜃0 + 𝛽𝑡

 System has no real eigenvectors the phase plane portrait is as the following
x'=-2x-3y x'=y
y'=3x-2y y'=-x-y

4 4

3 3

2 2

1 1

0 0

y
y

-1 -1

-2 -2

-3 -3

-4 -4

-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
x x

 The trajectories are spiral around the origin and toward the origin.

 This equilibrium point is called stable focus.

55
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Example 10: 𝑥+𝑥+𝑥 =0

0 1
𝑋= 𝑋
−1 −1
0 1
det 𝜆𝐼 − = 𝜆2 + 𝜆 + 1 = 0 𝜆1 , 𝜆2 = −0.5 ± 0.866𝑖
−1 −1
x'=y
y'=-x-y

4
2
3
x(0) = [1,-2]
1.5
2 X(0) = [1,2]
1
1

0 0.5
y

-1
0
-2
-0.5
-3
-1
-4 0 2 4 6 8 10
-4 -3 -2 -1 0 1 2 3 4
x

56
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

2. B ) 𝝀𝟐 , 𝝀𝟏 = 𝜶 ± 𝜷𝒊 , 𝜶 > 𝟎 ,𝜷 ≠ 𝟎 𝑟 𝑡 = 𝑟0 𝑒 𝛼𝑡

𝜃 𝑡 = 𝜃0 + 𝛽𝑡

 System has no real eigenvectors the phase plane portrait is as the following
x'=-2x-3y x'=y
y'=3x-2y y'=-x-y

4 4

3 3

2 2

1 1

0 0

y
y

-1 -1

-2 -2

-3 -3

-4 -4

-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
x x

 The trajectories are spiral around the origin and diverge from the origin.

 This equilibrium point is called unstable focus.

57
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Example 11: 𝑥−𝑥+𝑥 =0

0 1
𝑋= 𝑋
−1 1
0 1
det 𝜆𝐼 − = 𝜆2 − 𝜆 + 1 = 0 𝜆1 , 𝜆2 = 0.5 ± 0.866𝑖
−1 1
x'=y
y'=-x+y

4
200
3

2
0
1

0 -200
y

-1
X(0) = [1,2]
-2
-400
X(0) = [1,-2]
-3
-600
-4 0 2 4 6 8 10
-4 -3 -2 -1 0 1 2 3 4
x

58
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

2. C ) 𝝀𝟐 , 𝝀𝟏 = ±𝜷𝒊 , 𝜶 = 𝟎 ,𝜷 ≠ 𝟎 𝑟 𝑡 = 𝑟0 𝑒 𝛼𝑡

𝜃 𝑡 = 𝜃0 + 𝛽𝑡

 System has two imaginary eigenvalues and no real eigenvectors the phase
plane portrait is as the following
x'=3y x'=y
y'=-3x y'=-5x

4
4

3 3

2 2

1 1

0 0

y
y

-1 -1

-2 -2

-3 -3

-4 -4
-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
x x

 The trajectories are closed trajectories around the origin.

 This equilibrium point is marginally stable and is called center.

59
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Example 12: 𝑥 + 3𝑥 = 0
0 1
𝑋= 𝑋
−3 0
0 1
det 𝜆𝐼 − = 𝜆2 + 3 = 0 𝜆1 , 𝜆2 = ±1.732𝑖
−3 0
x'=y
y'=-3x

2
3

2
1
1 X(0) = [1,-2]
0 X(0) = [1,2]
0
y

-1
-1
-2

-3 -2
0 2 4 6 8 10
-4
-4 -3 -2 -1 0 1 2 3 4
x

60
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

3. Two similar (real) eigenvalues and two eigenvectors

𝜆 0
𝑦 = 𝐽𝑦 𝐽=
0 𝜆

𝑦1 = 𝜆𝑦1 𝑦1 = 𝑦10 𝑒 𝜆𝑡

𝑦2 = 𝜆𝑦2 𝑦2 = 𝑦20 𝑒 𝜆𝑡

𝑦1 𝑦10
= 𝑦2 = 𝐾𝑦1
𝑦2 𝑦20

61
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

3 ) 𝝀𝟐 = 𝝀𝟏 = 𝝀 ≠ 𝟎

 System has two similar eigenvalues and two different eigenvectors. The
phase plane portrait is as the following, depending to the sign of 𝜆
x'=2x x'=-2x
y'=2y y'=-2y

4 4

3 3

2 2

1 1

0 0

y
y

-1 -1

-2 -2

-3 -3

-4 -4

-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4

𝜆<0
x x

𝜆>0

 The trajectories are all along the initial conditions and they are 𝜆 < 0
toward 𝜆 > 0 or outward the origin
62
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

4. Two similar (real) eigenvalues and One eigenvectors

𝜆 1
𝑦 = 𝐽𝑦 𝐽=
0 𝜆

𝑦1 = 𝜆𝑦1 + 𝑦2 𝑦1 = 𝑦10 𝑒 𝜆𝑡 + 𝑦20 𝑡𝑒 𝜆𝑡

𝑦2 = 𝜆𝑦2 𝑦2 = 𝑦20 𝑒 𝜆𝑡

𝑦2 1 𝑦2
𝑦1 = 𝑦10 + 𝑦2 ln( )
𝑦20 𝜆 𝑦20

𝑦10 1 𝑦2
𝑦1 = 𝑦2 ( + ln )
𝑦20 𝜆 𝑦20

63
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

4 ) 𝝀𝟐 = 𝝀𝟏 = 𝝀 ≠ 𝟎

 System has two similar eigenvalues and only one eigenvector. The phase
plane portrait is as the following, depending to the sign of 𝜆
x ' = 0.5 x + y x ' = - 0.5 x + y
y ' = 0.5 y y ' = - 0.5 y

4 4

3 3

2 2

1
1

y
0
y

-1
-1

-2
-2

-3
-3

-4
-4
-4 -3 -2 -1 0 1 2 3 4

𝜆<0
-4 -3 -2 -1 0 1 2 3 4
x
x

𝜆>0

 The trajectories converge to zero or diverge to infinity along the system

eigenvector.
64
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems
3
X(0) = [2,3]
Example 13: 𝑥 + 2𝑥 + 𝑥 = 0 2 X(0) = [2,-4]

0 1
𝑋= 𝑋 0
−1 −2 -1
0 2 4 6 8 10

0 1
det 𝜆𝐼 − = 𝜆2 + 2𝜆 + 1 = 0 𝜆1 , 𝜆2 = −1
−1 −2

−1 −1 𝑣11
=0 2
1 −1 + 2 𝑣12
0
y

−0.71
𝑣1 =
0.71 -2

-4
-4 -2 0 2 4
x

65
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

A is singular (𝒅𝒆𝒕 𝑨 = 𝟎):

 System has at least one eigenvalue equal to zero and therefore infinite
number of equilibrium points. Three different categories can be
specified

 𝜆1 = 0 , 𝜆2 ≠ 0

 𝜆1 , 𝜆1 = 0 , 𝑅𝑎𝑛𝑘 𝐴 = 1

 𝜆1 , 𝜆1 = 0 , 𝑅𝑎𝑛𝑘 𝐴 = 0

66
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems
0 0
1) 𝜆1 = 0 , 𝜆2 ≠ 0 𝑦 = 𝐽𝑦 𝐽=
0 𝜆
𝑦1 = 0 𝑦1 = 𝑦10

x'=0
𝑦2 = 𝜆𝑦2
x'=-x+y
𝑦2 = 𝑦20 𝑒 𝜆𝑡
y'=2y y'=-3x+3y

4 4

3 3

2 2

1 1

0 0

y
y

-1 -1

-2 -2

-3 -3

-4 -4

-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
x x

 System has infinite number of non-isolated equilibrium points along a line

 System has two eigenvectors. Eigenvector corresponding to zero eigenvalue

is in fact loci of the equilibrium points

 Depending on the sign of the second eigenvalue, all the trajectories move
inward or outward to 𝑣1 along 𝑣2
67
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems
3
X(0) = [2,-4]

Example 14: 𝑥+𝑥 =0

2 X(0) = [3,-4]

0 1 0

𝑋= 𝑋
0 −1
-1

-2
0 2 4 6 8 10

0 1
det 𝜆𝐼 − = 𝜆2 + 𝜆 = 0 𝜆1 = 0 , 𝜆2 = −1
0 −1
0 −1 𝑣11 4
=0
0 1 𝑣12
2
1
𝑣1 =
0 0
y

−1 −1 𝑣11
0 −1 + 1 2 =0 -2
𝑣1
-4
1
𝑣1 = -4 -2 0 2 4
−1 x
68
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems
0 1
2) 𝜆1 , 𝜆1 = 0 , 𝑅𝑎𝑛𝑘 𝐴 = 1 𝑦 = 𝐽𝑦 𝐽=
0 0
𝑦1 = 𝑦2 𝑦1 = 𝑦20 𝑡 + 𝑦10

𝑦2 = 0 𝑦2 = 𝑦20

 System has infinite number of non-isolated equilibrium points along a line

 System has only one eigenvector, and it is loci of the equilibrium points

 All the trajectories move toward infinity along the system eigenvector
(unstable system).
69
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

2) 𝜆1 , 𝜆1 = 0 , 𝑅𝑎𝑛𝑘 𝐴 = 0

0 0
𝑦 = 𝐽𝑦 𝐽=
0 0

𝑦1 = 0 𝑦1 = 𝑦10

𝑦2 = 0 𝑦2 = 𝑦20

 System is a static system. All the points are equilibrium points

70
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Summary

Six different type of isolated equilibrium points can be identified

Stable/unstable node

Saddle point

Stable/ unstable focus

Center

71
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Stability Analysis of Higher Order Systems:

Analysis and results for the second order LTI system can be
extended to higher order LTI system

Graphical tool is not useful for higher order LTI system except
for third order systems.

This means stability analysis of mechanical system with more

than one DOF can not be materialized graphically

Stability analysis is performed through the eigenvalue analysis of

the A matrix.
72
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Consider a linear time invariant (LTI) system

x = Ax  Bu
y  Cx  Du
Origin is the only equilibrium point of the system if A is non-
singular
det( A )  0
x = Ax  xe  0

Otherwise the system has infinite number of equilibrium points,

all the points on null-space of A are in fact equil. points of the
system. det( A )  0
x = Ax  xe  x* | x*  Nullspace(A)

73
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Details for Case of Non-Singular A

 Origin is the only equilibrium point of the system

 This equilibrium point (system) is

 Exponentially stable if all eigenvalues of A are either real

negative or complex with negative real part.

 Marginally stable if eigenvalues of A have non-positive real

part and 𝑟𝑎𝑛𝑘 𝐴 − 𝜆𝐼 = 𝑛 − 𝑟 for all repeated imaginary
eigenvalues, 𝜆 with multiplicity of r

 Unstable, otherwise.

74
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems
A is Non-Singular

 Classification of the equilibrium point of higher order system

into node, focus, and saddle point is not as easy as second order
system. However some points can be emphasized:

 The equilibrium point is stable/unstable node if all

eigenvalues are real and have the negative/positive sign.

 The equilibrium point is center if a pair of eigenvalues are

pure imaginary complex conjugate and all other eigenvalues
have negative real

 In the case of different sign in the real part of the

eigenvalues trajectories have the saddle type behavior near
the equilibrium point
75
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Trajectories are along the eigenvector with minimum

absolute real part near the equilibrium point and along the
eigenvector with maximum absolute real part.

Trajectories have spiral behavior if there exist some complex

(obviously conjugates) eigenvalues.

Spiral behavior is toward/outward depending on the sign

(negative and positive) of real part of the complex conjugate
eigenvalues.

These concepts can be visualized and better understood in three

dimensional case

76
Advanced Control (Mehdi Keshmiri, Winter 95)
Lyapunov Indirect Method in

Stability Analysis of

Nonlinear Systems

77
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

 There are two conventional approaches in the stability analysis

of nonlinear systems:

 Lyapunov direct method

 Lyapunov indirect method or linearization approach

 The direct method analyzes stability of the system (equilibrium

point) using the nonlinear equations of the system

 The indirect method analyzes the system stability using the

linearized equations about the equilibrium point.

78
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems
Motivation:
A nonlinear system near its equilibrium point behaves like a linear:

• Nonlinear system: x  f (x)

• Equilibrium point: f (x)  0  xe

• Motion about equilibrium point: x  xe  xˆ

• Linearized motion:
f
x  f (x)  xˆ  f (xe  xˆ )  f (xe )  xˆ  H.O.T  xˆ  Axˆ
x x e

if xe  0
• It means near 𝑥𝑒 :
x  f ( x) xˆ  Axˆ  x  xˆ
79
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

This means stability of the equilibrium point may be studied

through the stability analysis of the linearized system.

This is the base of the Lyapunov Indirect Method

Example: in the nonlinear second order system

x1  x22  x1 cos x2
x2  x2  (1  x1 ) x1  x1 sin x2
origin is the equilibrium point and the linearized system is given
by
 xˆ1   f   xˆ1   xˆ1  xˆ1
    
 xˆ2   x  xe 0  xˆ2   xˆ2  xˆ1  xˆ2
80
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Theorem (Lyapunov Linearization Method):

 If the linearized system is strictly stable (i.e. all eigenvalues of A

are strictly in the left half complex plane ) then the equilibrium
point in the original nonlinear system is asymptotically stable.

xˆ  Axˆ isstrictlystable  x  f (x) is asymptoticallystable

 If the linearized system is unstable (i.e. in the case of right half

plane eigenvalue(s) or repeated eigenvalues on the imaginary axis
with geometrical deficiency (𝑟 > 𝑛 − 𝑟𝑎𝑛𝑘(𝜆𝐼 − 𝐴)), then the
equilibrium point in the original nonlinear system is unstable.
xˆ  Axˆ is unstable  x  f (x) is unstable
81
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Theorem (Lyapunov Linearization Method):

 If the linearized system is marginally stable (i.e. all eigenvalues of A are in the
left half complex plane and eigenvalues on the imaginary axis have no
geometrical deficiency) then one cannot conclude anything from the linear
approximation. The equilibrium point in the original nonlinear system may
be stable, asymptotically stable, or unstable.

x  f (x) is asymptoticallystable

xˆ  Axˆ is marginallystable  x  f (x) is marginallystable
x  f (x) is unstable


 The Lyapunov linearized approximation method only talks about the local
stability of the nonlinear system, if anything can be concluded.

82
Advanced Control (Mehdi Keshmiri, Winter 95)
 Phase Plane Analysis of LTI Systems

Example 15:

 The nonlinear system 𝑥 = 𝑎𝑥 + 𝑏𝑥 5 is

 Asymptotically stable if 𝑎 < 0

 Unstable if 𝑎 > 0

 No conclusion from linear approximation can be drawn if

 The origin in the nonlinear second order system

x1  x22  x1 cos x2
x2  x2  (1  x1 ) x1  x1 sin x2
 xˆ1  1 0  xˆ1 
 is unstable because the linearized system      is
unstable  xˆ2  1 1   xˆ2 
83
Advanced Control (Mehdi Keshmiri, Winter 95)