This action might not be possible to undo. Are you sure you want to continue?

BooksAudiobooksComicsSheet Music### Categories

### Categories

### Categories

Editors' Picks Books

Hand-picked favorites from

our editors

our editors

Editors' Picks Audiobooks

Hand-picked favorites from

our editors

our editors

Editors' Picks Comics

Hand-picked favorites from

our editors

our editors

Editors' Picks Sheet Music

Hand-picked favorites from

our editors

our editors

Top Books

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Audiobooks

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Comics

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Sheet Music

What's trending, bestsellers,

award-winners & more

award-winners & more

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

Nonlinear Control

Lecture 2:Phase Plane Analysis

Farzaneh Abdollahi

Department of Electrical Engineering

Amirkabir University of Technology

Fall 2010

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 1/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

**Phase Plane Analysis: is a graphical
**

method for studying second-order systems

by

**providing motion trajectories
**

corresponding to various initial

conditions.

**then examine the qualitative features of
**

the trajectories.

**ﬁnally obtaining information regarding
**

the stability and other motion patterns

of the system.

**It was introduced by mathematicians such
**

as Henri Poincare in 19th century.

http://en.wikipedia.org/wiki/Henri_Poincar%C3%A9

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 2/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Motivations

Importance of Knowing Phase Plane Analysis:

**Since it is on second-order, the solution trajectories can be
**

represented by carves in plane provides easy visualization of the

system qualitative behavior.

**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.

**Disadvantage of Phase Plane Method: It is restricted to at most
**

second-order and graphical study of higher-order is computationally

and geometrically complex.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 3/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Introduction

Motivation

Phase Plane

Vector Field Diagram

Isocline Method

Qualitative Behavior of Linear Systems

Case 1: λ

1

= λ

2

= 0

Case 2: Complex Eigenvalues, λ

1,2

= α ±j β

Case 3: Nonzero Multiple Eigenvalues λ

1

= λ

2

= λ = 0

Case 4: Zero Eigenvalues

Local Behavior of Nonlinear Systems

Limit Cycle

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 4/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Concept of Phase Plane

**Phase plane method is applied to autonomous 2nd order system
**

described as follows:

˙ x

1

= f

1

(x

1

, x

2

) (1)

˙ x

2

= f

2

(x

1

, x

2

) (2)

f

1

, f

2

: R

2

→R.

System response (x(t) = (x

1

(t), x

2

(t))) to initial condition

x

0

= (x

10

, x

20

) is a mapping from R to R

2

.

The x

1

−x

2

plane is called State plane or Phase plane

**The locus in the x
**

1

−x

2

plane of the solution x(t) for all t ≥ 0 is a

curve named trajectory or orbit that passes through the point x

0

**The family of phase plane trajectories corresponding to various initial
**

conditions is called Phase protrait of the system.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 5/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

How to Construct Phase Plane Trajectories?

**Despite of exiting several routines to
**

generate the phase portraits by computer, it

is useful to learn roughly sketch the portraits

or quickly verify the computer outputs.

**Some methods named: Isocline, Vector ﬁeld
**

diagram, delta method, Pell’s method, etc

Vector Field Diagram:

Revisiting (1) and (2):

˙ x = f (x) =

_

f

1

(x)

f

2

(x)

_

, ˙ x = ( ˙ x

1

, ˙ x

2

)

To each vector (x

1

, x

2

), a corresponding

vector (f

1

(x

1

, x

2

), f

2

(x

1

, x

2

)) known as a

vector ﬁeld is associated.

Example: If f (x) = (2x

2

1

, x

2

), for

x = (1, 1), next point is

(1, 1) + (2, 1) = (3, 2)

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 6/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Vector Field Diagram

**By repeating this for suﬃcient point
**

in the state space, a vector ﬁeld

diagram is obtained.

Noting that

dx

2

dx

1

=

f

2

f

1

vector ﬁeld at

a point is tangent to trajectory

through that point.

∴ starting from x

0

and by using the

vector ﬁeld with suﬃcient points,

the trajectory can be constructed.

**Example: Pendulum without friction
**

˙ x

1

= x

2

˙ x

2

= −10 sin x

1

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 7/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Isocline Method

The term isocline derives from the Greek words for ”same slope.”

**Consider again Eqs (1) and (2), the slope of the trajectory at point x:
**

S(x) =

dx

2

dx

1

=

f

2

(x

1

, x

2

)

f

1

(x

1

, x

2

)

An isocline with slope α is deﬁned as S(x) = α

**∴ all the points on the curve f
**

2

(x

1

, x

2

) = αf

1

(x

1

, x

2

) have the same

tangent slope α.

**Note that the ”time” is eliminated here The responses x
**

1

(t) and x

2

(t)

cannot be obtained directly.

**Only qualitative behavior can be concluded, such as stable or oscillatory
**

response.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 8/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Isocline Method

**The algorithm of constructing the phase portrait by isocline method:
**

1. Plot the curve S(x) = α in state-space (phase plane)

2. Draw small line with slope α. Note that the direction of the line depends on

the sign of f

1

and f

2

at that point.

3. Repeat the process for suﬃcient number of α s.t. the phase plane is full of

isoclines.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 9/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Example: Pendulum without Friction

**Consider the dynamics ˙ x
**

1

= x

2

, ˙ x

2

= −sinx

1

∴S(x) =

−sinx

1

x

2

= c

Isoclines: x

2

=

−1

c

sinx

1

**Trajectories for diﬀerent init. conditions can be obtained by using the
**

given algorithm

**The response for x
**

0

= (

π

2

, 0) is depicted in Fig.

**The closed curve trajectory conﬁrms marginal stability of the system.
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 10/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Example: Pendulum with Friction

**Dynamics of pendulum with friction:
**

˙ x

1

= x

2

, ˙ x

2

= −0.5x

2

−sinx

1

∴S(x) =

−0.5−sinx

1

x

2

= c

Isoclines: x

2

=

−1

0.5+c

sinx

1

Similar Isoclines but with diﬀerent slopes

**Trajectory is drawn for x
**

0

= (

π

2

, 0)

**The trajectory shrinks like an spiral converging to the origin
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 11/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Qualitative Behavior of Linear Systems

**First we analyze the phase plane of linear systems since the behavior of
**

nonlinear systems around equilibrium points is similar of linear ones

**For LTI system:
**

˙ x = Ax, A ∈ R

2×2

, x

0

: initial statex(t) = Me

J

r

t

M

−1

x

0

J

r

: Jordan block of A, M : Matrix of eigenvectors M

−1

AM = J

r

**Depending on the eigenvalues of A, J
**

r

has one of the following forms:

λ

i

: real & distinct J

r

=

_

λ

1

0

0 λ

2

_

λ

i

: real & multiple J

r

=

_

λ k

0 λ

_

, k = 0, 1,

λ

i

: complex J

r

=

_

α −β

β α

_

**The system behavior is diﬀerent at each case
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 12/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 1: λ

1

= λ

2

= 0

In this case M = [v

1

v

2

] where v

1

and v

2

are real eigenvectors associated

with λ

1

and λ

2

**To transform the system into two decoupled ﬁrst-order diﬀ equations, let
**

z = M

−1

x:

˙ z

1

= λ

1

z

1

˙ z

2

= λ

2

z

2

**The solution for initial states (z
**

01

, z

02

):

z

1

(t) = z

10

e

λ

1

t

, z

2

(t) = z

20

e

λ

2

t

eliminating tz

2

= Cz

λ

2

/λ

1

1

, C = z

20

/(z

10

)

λ

2

/λ

1

(3)

Phase portrait is obtained by changing C ∈ R and plotting (3).

**The phase portrait depends on the sign of λ
**

1

and λ

2

.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 13/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 1.1: λ

2

< λ

1

< 0

t →∞ ⇒ the terms e

λ

1

t

and e

λ

2

t

tend to zero

**Trajectories from entire state-space tend to
**

origin the equilibrium point x = 0 is stable

node.

e

λ

2

t

→0 faster λ

2

is fast eignevalue and

v

2

is fast eigenvector.

**Slope of the curves:
**

dz

2

dz

1

= C

λ

2

λ

1

z

(λ

2

/λ

1

−1)

1

λ

2

< λ

1

< 0λ

2

/λ

1

> 1, so slope is

zero as z

1

−→0

inﬁnity as z

1

−→∞.

∴ The trajectories are

tangent to z

1

axis, as they approach to origin

parallel to z

2

axis, as they are far from origin.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 14/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 1.1: λ

2

< λ

1

< 0

Since z

2

approaches to zero faster than z

1

, trajectories are sliding

along z

1

axis

In X plane also trajectories are:

**tangent to the slow eigenvector v
**

1

for near origin

**parallel to the fast eigenvector v
**

2

for far from origin

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 15/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 1.2: λ

2

> λ

1

> 0

t →∞ ⇒ the terms e

λ

1

t

and e

λ

2

t

grow exponentially, so

The shape of the trajectories are the same, with opposite directions

**The equilibrium point is socalled unstable node
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 16/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 1.3: λ

2

< 0 < λ

1

t →∞ ⇒ e

λ

2

t

−→0, but e

λ

1

t

−→∞,so

λ

2

: stable eigenvalue, v

2

: stable eigenvector

λ

1

: unstable eigenvalue, v

1

: unstable eigenvector

**Trajectories are negative exponentials since
**

λ

2

λ

1

is negative.

Trajectories are

decreasing in z

2

direction, but increasing in z

1

direction

tangent to z

1

as |z

1

| →∞ and tangent to z

2

as |z

1

| →0

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 17/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 1.3: λ

2

< 0 < λ

1

The exceptions of this hyperbolic shape:

**two trajectories along z
**

2

-axis →0 as t →0, called stable trajectories

**two trajectories along z
**

1

-axis →∞ as t →0, called unstable trajectories

This equilibrium point is called saddle point

**Similarly in X plane, stable trajectories are along v
**

2

, but unstable

trajectories are along the v

1

For λ

1

< 0 < λ

2

the direction of the trajectories are changed.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 18/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 2: Complex Eigenvalues, λ

1,2

= α ±j β

˙ z

1

= αz

1

−βz

2

˙ z

2

= βz

1

+αz

2

**The solution is oscillatory =⇒ polar coordinates
**

(r =

_

z

2

1

+ z

2

2

, θ = tan

−1

(

z

2

z

1

))

˙ r = αr r (t) = r

0

e

αt

˙

θ = βθ(t) = θ

0

+βt

**This results in Z plane is a logarithmic spiral where α determines the form
**

of the trajectories:

**α < 0 : as t →∞r →0 and angle θ is rotating. The spiral converges to
**

origin =⇒ Stable Focus.

**α > 0: as t →∞r →∞ and angle θ is rotating. The spiral diverges
**

away from origin =⇒ Unstable Focus.

**α = 0: Trajectories are circles with radius r
**

0

=⇒ Center

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 19/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 2: Complex Eigenvalues, λ

1,2

= α ±j β

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 20/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 3: Nonzero Multiple Eigenvalues λ

1

= λ

2

= λ = 0

Let z = M

−1

x:

˙ z

1

= λz

1

+ kz

2

, ˙ z

2

= λz

2

the solution is :z

1

(t) = e

λt

(z

10

+ kz

20

t), z

2

(t) = z

20

e

λt

z

1

= z

2

_

z

10

z

20

+

k

λ

ln

_

z

2

z

20

__

Phase portrait are depicted for k = 0 and k = 1.

When the eignevectors are diﬀerent k = 0:

similar to Case 1, for λ < 0 is stable, λ > 0 is unstable.

**Decaying rate is the same for both modes (λ
**

1

= λ

2

) trajectories are lines

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 21/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 3: Nonzero Multiple Eigenvalues λ

1

= λ

2

= λ = 0

There is no fast-slow asymptote.

**k = 1 is more complex, but it is still similar to Case 1:
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 22/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 4.1: One eigenvalue is zero λ

1

= 0, λ

2

= 0

A is singular in this case

**Every vector in null space of A is an
**

equilibrium point

**There is a line (subspace) of equilibrium
**

points

M = [v

1

v

2

] , v

1

, v

2

: corresponding

eigenvectors, v

1

∈ N(A).

˙ z

1

= 0, ˙ z

2

= λ

2

z

2

solution: z

1

(t) = z

10

, z

2

(t) = z

20

e

λ

2

t

**Phase portrait depends on sign of λ
**

2

:

λ

2

< 0: Trajectories converge to

equilibrium line

λ

2

> 0: Trajectories diverge from

equilibrium line

**in X plane, the equilibrium set is v
**

1

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 23/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Case 4.2: Both eigenvalues zero λ

1

= λ

2

= 0

Let z = M

−1

x

˙ z

1

= z

2

, ˙ z

2

= 0

solution: z

1

(t) = z

10

+ z

20

t, z

2

(t) = z

20

z

1

linearly increases/decreases base on the sign of z

20

z

2

axis is equilibrium subspace in Z-plane

Dotted line is equilibrium subspace

**The diﬀerence between Case 4.1 and 4.2: all trajectories start oﬀ the
**

equilibrium set move parallel to it.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 24/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

As Summary:

Six types of equilibrium points can be identiﬁed:

stable/unstable node

saddle point

stable/ unstable focus

center

Type of equilibrium point depends on sign of the eigenvalues

If real part of eignevalues are Positive unstability

If real part of eignevalues are Negative stability

All properties for linear systems hold globally

**Properties for nonlinear systems only hold locally
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 25/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Local Behavior of Nonlinear Systems

**Qualitative behavior of nonlinear systems is obtained locally by
**

linearization around the equilibrium points

**Type of the perturbations and reaction of the system to them determines
**

the degree of validity of this analysis

**A simple example: Consider the linear perturbation case
**

A −→A + ∆A, where ∆A ∈ R

2×2

: small perturbation

Eigenvalues of a matrix continuously depend on its parameters

**Positive (Negative) eigenvalues of A remain positive (negative) under small
**

perturbations.

**For eigenvalues on the j ω axis no matter how small perturbation is, it
**

changes the sign of eigenvalue.

Therefore

**node or saddle point or focus equilibrium point remains the same under
**

small perturbations

**This analysis is not valid for a center equilibrium point
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 26/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Multiple Equilibria

Linear systems can have

an isolated equilibrium point or

a continuum of equilibrium points (When detA = 0)

**Unlike linear systems, nonlinear systems can have multiple isolated
**

equilibria.

**Qualitative behavior of second-order nonlinear system can be
**

investigated by

generating phase portrait of system globally by computer programs

**linearize the system around equilibria and study the system behavior near
**

them without drawing the phase portrait

Let (x

10

, x

20

) are equilibrium points of

˙ x

1

= f

1

(x

1

, x

2

)

˙ x

2

= f

2

(x

1

, x

2

) (4)

f

1

, f

2

are continuously diﬀerentiable about (x

10

, x

20

)

**Since we are interested in trajectories near (x
**

10

, x

20

), deﬁne

x

1

= y

1

+ x

10

, x

2

= y

2

+ x

20

y

1

, y

2

are small perturbations form equilibrium point.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 27/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Qualitative Behavior Near Equilibrium Points

**Expanding (4) into its Taylor series
**

˙ x

1

= ˙ x

10

+ ˙ y

1

= f

1

(x

10

, x

20

)

. ¸¸ .

0

+

∂f

1

∂x

1

¸

¸

¸

¸

(x

10

,x

20

)

y

1

+

∂f

1

∂x

2

¸

¸

¸

¸

(x

10

,x

20

)

y

2

+ H.O.T.

˙ x

2

= ˙ x

20

+ ˙ y

2

= f

2

(x

10

, x

20

)

. ¸¸ .

0

+

∂f

2

∂x

1

¸

¸

¸

¸

(x

10

,x

20

)

y

1

+

∂f

2

∂x

2

¸

¸

¸

¸

(x

10

,x

20

)

y

2

+ H.O.T.

**For suﬃciently small neighborhood of equilibrium points, H.O.T. are
**

negligible

_

˙ y

1

= a

11

y

1

+ a

12

y

2

˙ y

2

= a

21

y

1

+ a

22

y

2

, a

ij

=

∂f

i

∂x

¸

¸

¸

¸

x

0

, i = 1, 2

**The equilibrium point of the linear system is (y
**

1

= y

2

= 0)

˙ y = Ay, A =

_

a

11

a

12

a

21

a

22

_

=

_

∂f

1

∂x

1

∂f

1

∂x

2

∂f

2

∂x

1

∂f

2

∂x

2

_

=

∂f

∂x

¸

¸

¸

¸

x

0

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 28/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Qualitative Behavior Near Equilibrium Points

Matrix

∂f

∂x

is called Jacobian Matrix.

**The trajectories of the nonlinear system in a small neighborhood of an
**

equilibrium point are close to the trajectories of its linearization about

that point:

if the origin of the linearized state equation is a

stable (unstable) node, or a stable (unstable) focus or a saddle point,

**then in a small neighborhood of the equilibrium point, the trajectory of
**

the nonlinear system will behave like a

**stable (unstable) node, or a stable (unstable) focus or a saddle point.
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 29/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Example: Tunnel Diode Circuit

˙ x

1

=

1

C

[−h(x

1

) + x

2

]

˙ x

2

=

1

L

[−x

1

−Rx

2

+ u]

**u = 1.2v, R = 1.5KΩ, C = 2pF, L = 5µH, time in nanosecond, current
**

in mA

˙ x

1

= 0.5[−h(x

1

) + x

2

]

˙ x

2

= 0.2[−x

1

−1.5x

2

+ 1.2]

Suppose h(x

1

) = 17.76x

1

−103.79x

2

1

+ 229.62x

3

1

−226.31x

4

1

+ 83.72x

5

1

equilibrium points ( ˙ x

1

= ˙ x

2

= 0):

Q

1

= (0.063, 0.758), Q

2

= (0.285, 0.61), Q

3

= (0.884, 0.21)

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 30/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Example: Tunnel Diode Circuit

**The global phase portrait is generated by
**

a computer program is shown in Fig.

**Except for two special trajectories which
**

approach Q

2

, all trajectories approach

either Q

1

or Q

3

.

**Near equilibrium points Q
**

1

and Q

3

are

stable nodes,Q

2

is like saddle point.

**The two special trajectories from a curve
**

that divides the plane into two halves with

diﬀerent behavior (separatrix curves).

**All trajectories originating from left side
**

of the curve approach to Q

1

**All trajectories originating from left side
**

of the curve approach to Q

3

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 31/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Tunnel Diode:Qualitative Behavior Near Equilibrium Points

Jacobian matrix

∂f

∂x

=

_

0.5

´

h(x

1

) 0.5

−0.2 −0.3

_

´

h(x

1

) =

dh

dx

1

= 17.76 −207.58x

1

+ 688.86x

2

1

−905.24x

3

1

+ 418.6x

4

1

**Evaluate the Jacobian matrix at the equilibriums Q
**

1

, Q

2

, Q

3

:

Q

1

= (0.063, 0.758), A

1

=

_

−3.598 0.5

−0.2 −0.3

_

, λ

1

= −3.57, λ

2

= −0.33 stable node

Q

2

= (0.285, 0.61), A

2

=

_

1.82 0.5

−0.2 −0.3

_

, λ

1

= 1.77, λ

2

= −0.25 saddle point

Q

3

= (0.884, 0.21), A

3

=

_

−1.427 0.5

−0.2 −0.3

_

, λ

1

= −1.33, λ

2

= −0.4 stable node

**∴ similar results given from global phase portrait.
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 32/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Tunnel Diode Circuit

**In practice, There are only two stable equilibrium points: Q
**

1

or Q

3

.

Equilibrium point at Q

2

in never observed,

**Even if set up the exact initial conditions corresponding t Q
**

2

, the

ever-present physical noise causes the trajectory to diverge from Q

2

**Such circuit is called bistable, since it has two steady-state operating
**

points.

Triggering formQ

1

to Q

3

or vice versa is achieved by changing the load

line

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 33/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

**Special case: If the Jacobian matrix has eigenvalues on j ω, then the
**

qualitative behavior of nonlinear system near the equilibrium point could

be quite distinct from the linearized one.

Example:

˙ x

1

= −x

2

−µx

1

(x

2

1

+ x

2

2

)

˙ x

2

= x

1

−µx

2

(x

2

1

+ x

2

2

)

**It has equilibrium point at origin x
**

∗

= 0.

A =

_

0 −1

1 0

_

⇒λ

1,2

= ±j ⇒ center

**Now consider the nonlinear system
**

x

1

= r cos θ, x

2

= r sin θ ⇒ ˙ r = −µr

3

,

˙

θ = 1

∴ nonlinear system is stable if µ > 0 and is unstable if µ < 0

**∴ the qualitative behavior of nonlinear and linearized one are diﬀerent.
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 34/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Limit Cycle

**A system oscillates when it has a nontrivial periodic solution
**

x(t + T) = x(t), ∀t ≥ 0, for someT > 0

The word ”nontrivial” is used to exclude the constant solutions.

**The image of a periodic solution in the phase portrait is a closed
**

trajectory, calling periodic orbit or closed orbit.

We have already seen oscillation of linear system with eigenvalues ±j β.

The origin of the system is a center, and the trajectories are closed

**the solution in Jordan form:
**

z

1

(t) = r

0

cos(βt +θ

0

), z

2

= r

0

sin(βt +θ

0

)

r

0

=

_

z

2

10

+ z

2

20

, θ

0

= tan

−1

z

20

z

10

r

0

: amplitude of oscillation

**Such oscillation where there is a continuum of closed orbits is referred to
**

harmonic oscillator.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 35/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Limit Cycle

**The physical mechanism leading to these oscillations is a periodic exchange of
**

energy stored in the capacitor (electric ﬁeld) and the inductor (magnetic ﬁeld).

**We have seen that such oscillation is not robust any small perturbations
**

destroy the oscillation.

The linear oscillator is not structurally stable

The amplitude of the oscillation depends on the initial conditions.

**These problems can be eliminated in nonlinear oscillators. A practical nonlinear
**

oscillator can be build such that

The nonlinear oscillator is structurally stable

**The amplitude of oscillation (at steady state) is independent of initial
**

conditions.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 36/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Limit Cycle

On phase plane, a limit cycle is deﬁned as an isolated closed orbit.

**For limit cycle the trajectory should be
**

1. closed: indicating the periodic nature of the motion

2. isolated: indicating limiting nature of the cycle with nearby trajectories

converging to/ diverging from it.

The mass spring damper does not have limit cycle; they are not isolated.

**Depends on trajectories motion pattern in vicinity of limit cycles, there
**

are three type of limit cycle:

**Stable Limit Cycles: as t →∞ all trajectories in the vicinity converge to
**

the limit cycle.

**Unstable Limit Cycles: as t →∞ all trajectories in the vicinity diverge from
**

the limit cycle.

**Semi-stable Limit Cycles: as t →∞ some trajectories in the vicinity
**

converge to/ and some diverge from the limit cycle.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 37/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Limit Cycle

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 38/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Example1.a: stable limit cycle

˙ x

1

= x

2

−x

1

(x

2

1

+ x

2

2

−1)

˙ x

2

= −x

1

−x

2

(x

2

1

+ x

2

2

−1)

Polar coordinates (x

1

:= rcos(θ), x

2

=: rsin(θ))

˙ r = −r (r

2

−1)

˙

θ = −1

**If trajectories start on the unit circle (x
**

2

1

(0) + x

2

2

(0) = r

2

= 1), then

˙ r = 0 =⇒ The trajectory will circle the origin of the phase plane with

period of

1

2π

.

r < 1 =⇒ ˙ r > 0 =⇒ trajectories converges to the unit circle from inside.

**r > 1 =⇒ ˙ r < 0 =⇒ trajectories converges to the unit circle from
**

outside.

**Unit circle is a stable limit cycle for this system.
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 39/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Example1.b: unstable limit cycle

˙ x

1

= x

2

+ x

1

(x

2

1

+ x

2

2

−1)

˙ x

2

= −x

1

+ x

2

(x

2

1

+ x

2

2

−1)

Polar coordinates (x

1

:= rcos(θ), x

2

=: rsin(θ))

˙ r = r (r

2

−1)

˙

θ = −1

**If trajectories start on the unit circle (x
**

2

1

(0) + x

2

2

(0) = r

2

= 1), then

˙ r = 0 =⇒ The trajectory will circle the origin of the phase plane with

period of

1

2π

.

**r < 1 =⇒ ˙ r < 0 =⇒ trajectories diverges from the unit circle from
**

inside.

**r > 1 =⇒ ˙ r > 0 =⇒ trajectories diverges from the unit circle from
**

outside.

**Unit circle is an unstable limit cycle for this system.
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 40/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Example1.c: semi stable limit cycle

˙ x

1

= x

2

−x

1

(x

2

1

+ x

2

2

−1)

2

˙ x

2

= −x

1

−x

2

(x

2

1

+ x

2

2

−1)

2

Polar coordinates (x

1

:= rcos(θ), x

2

=: rsin(θ))

˙ r = −r (r

2

−1)

2

˙

θ = −1

**If trajectories start on the unit circle (x
**

2

1

(0) + x

2

2

(0) = r

2

= 1), then

˙ r = 0 =⇒ The trajectory will circle the origin of the phase plane with

period of

1

2π

.

**r < 1 =⇒ ˙ r < 0 =⇒ trajectories diverges from the unit circle from
**

inside.

**r > 1 =⇒ ˙ r < 0 =⇒ trajectories converges to the unit circle from
**

outside.

**Unit circle is a semi-stable limit cycle for this system.
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 41/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Bendixson’s Criterion: Nonexistence Theorem of Limit Cycle

Gives a suﬃcient condition for nonexistence of a periodic solution:

**Suppose Ω is simply connected region in 2-dimention space in this region
**

we deﬁne ∇f =

∂f

1

∂x

1

+

∂f

2

∂x

2

. If ∇f is not identically zero over any

subregion of Ω and does not change sign in Ω, then Ω contain no limit

cycle for the nonlinear system

˙ x

1

= f

1

(x

1

, x

2

)

˙ x

2

= f

2

(x

1

, x

2

)

**Simply connected set: the boundary of the set is connected + the set is
**

connected

**Connected set: for connecting any two points belong to the set, there is a
**

line which remains in the set.

**The boundary of the set is connected if for connecting any two points
**

belong to boundary of the set there is a line which does not cross the set

exmaple of connected but not simply connected set

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 42/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Bendixson’s Criterion

Proof by contradiction:

Recall that

dx

2

dx

1

=

f

2

f

1

=⇒f

2

dx

1

−f

1

dx

2

= 0,

**∴ Along a closed curve L of a limit cycle:
**

_

L

(f

2

dx

1

−f

1

dx

2

) = 0

**Using Stoke’s Theorem: (
**

_

L

f .ndl =

_ _

S

∇fds = 0 S is enclosed by L)

_ _

S

_

∂f

1

∂x

1

+

∂f

2

∂x

2

_

dx

1

dx

2

= 0

This is true if

∇f = 0 ∀x ∈ S or

∇f changes sign in S

**This is in contradiction with the assumption that
**

∂f

1

∂x

1

+

∂f

2

∂x

2

does not vanish

and does not change sign there is no closed trajectory.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 43/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Nonexistence Theorem of Periodic Solutions for Linear

Systems

**Suﬃcient condition for nonexistence of a periodic solution in linear
**

systems:

˙ x

1

= a

11

x

1

+ a

12

x

2

˙ x

2

= a

21

x

1

+ a

22

x

2

∴

∂f

1

∂x

1

+

∂f

2

∂x

2

= a

11

+ a

22

= 0=⇒ no periodic sol.

**This is consistent with eigenvalue analysis form of center point which is
**

obtained for periodic solutions:

λ

2

−(a

11

+ a

22

) λ + (a

11

a

22

−a

12

a

21

) = 0

center ∴ a

11

+ a

22

= 0, a

11

a

22

−a

12

a

21

> 0

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 44/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Limit Cycle

**Example for nonexistence of limit cycle
**

˙ x

1

= g(x

2

) + 4x

1

x

2

2

˙ x

2

= h(x

1

) + 4x

2

1

x

2

∴

∂f

1

∂x

1

+

∂f

2

∂x

2

= 4(x

2

1

+ x

2

2

) > 0 ∀x ∈ R

2

**No limit cycle exist in R
**

2

for this system.

**Note that: there is no equivalent theorem for higher order systems.
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 45/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Poincare-Bendixson Criterion: Existence Theorem of Limit Cycle

**If there exists a closed and bounded set M s.t.
**

1. M contains no equilibrium point or contains only one equilibrium point such

that the Jacobian matrix

∂f

∂x

at this point has eigenvalues with positive real

parts (unstable focus or node).

2. Every trajectory starting in M stays in M for all future time

=⇒ M contains a periodic solution

**The idea behind the theorem is that all possible shape of limit points in a
**

plane (R

2

) are either equilibrium points or periodic solutions.

**Hence, if the positive limit set contains no equilibrium point, it must have
**

a periodic solution.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 46/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Poincare-Bendixson Criterion: Existence Theorem of Limit Cycle

**If M has unstable node/focus, in vicinity
**

of that equilibrium point all trajectories

move away

**By excluding the vicinity of unstable
**

node/focus, the set M is free of

equilibrium and all trajectories are trapped

in it.

**No equivalent theorem for R
**

n

, n ≥ 3.

**A solution could be bounded in R
**

3

, but

neither it is periodic nor it tends to a

periodic solution.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 47/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Example for Existence Theorem of Limit Cycle

˙ x

1

= −x

2

−x

1

(x

2

1

+ x

2

2

−1)

˙ x

2

= x

1

−x

2

(x

2

1

+ x

2

2

−1)

Polar coordinate:

˙ r = (1 −r

2

)r

˙

θ = −1

˙ r ≤ 0 for r ≥ 1 +η

1

, η

1

> 0

˙ r ≥ 0for r ≤ 1 −η

2

, 1 > η

2

> 0

**The area found by the circles with radius
**

1 −η

2

and 1 +η

1

satisﬁes the condition

of the P.B. theorem a periodic solution

exists.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 48/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Existence Theorem of Limit Cycle

A method to investigate whether or not trajectories remain inside M:

**Consider a simple closed curve V(x) = c, where V(x) is a continuously
**

diﬀerentiable function

The vector f at a point x on the curve points

**inward if the inner product of f and the gradient vector ∇V(x) is negative:
**

f (x).∇V(x) =

∂V

∂x

1

f

1

(x) +

∂V

∂x

2

f

2

(x) < 0

outward if f (x).∇V(x) > 0

tangent to the curve if f (x).∇V(x) = 0.

**Trajectories can leave a set only if the vector ﬁled points outward at some
**

points on the boundary.

**For a set of the form M = {V(x) ≤ c}, for some c > 0, trajectories
**

trapped inside M if f (x).∇V(x) ≤ 0 on the boundary of M.

**For an annular region of the form M = {W(x) ≥ c
**

1

and V(x) ≤ c

2

} for

some c

1

, c

2

> 0, trajectories remain inside M if f (x).∇V(x) ≤ 0 on

V(x) = c

2

and f (x).∇W(x) ≥ 0 on W(x) = c

1

.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 49/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Example for Existence Theorem of Limit Cycle

**Consider the system
**

˙ x

1

= x

1

+ x

2

−x

1

(x

2

1

+ x

2

2

)

˙ x

2

= −2x

1

+ x

2

−x

2

(x

2

1

+ x

2

2

)

The system has a unique equilibrium point at the origin.

**The Jacobian matrix
**

∂f

∂x

¸

¸

¸

¸

x=0

=

_

1 −3x

2

1

−x

2

2

1 −2x

1

x

2

−2 −2x

1

x

2

1 −x

2

1

−3x

2

2

_

x=0

=

_

1 1

−2 1

_

has eigenvalues at 1 ±j

√

2

Let M = V(x) ≤ c, where V(x) = x

2

1

+ x

2

2

.

**M is bounded and contains eigenvalues with positive real parts
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 50/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Example for Existence Theorem of Limit Cycle

**On the surface V(x) = c, we have
**

∂V

∂x

1

f

1

(x) +

∂V

∂x

2

f

2

(x) = 2x

1

[x

1

+ x

2

−x

1

(x

2

1

+ x

2

2

)]

+2x

2

[−2x

1

+ x

2

−x

2

(x

2

1

+ x

2

2

)]

= 2(x

2

1

+ x

2

2

) −2(x

2

1

+ x

2

2

)

2

−2x

1

x

2

≤ 2(x

2

1

+ x

2

2

) −2(x

2

1

+ x

2

2

)

2

+ (x

2

1

+ x

2

2

) = 3c −2c

2

Choosing c > 1.5 ensures that all trajectories trapped inside M.

**∴ by PB criterion, there exits at least one periodic orbit.
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 51/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Index Theorem

Let C be a simple closed curve not passing through any equilibrium point

Consider the orientation of the vector ﬁeld f (x) at a point p ∈ C.

**Let p traverse C counterclockwise the vector f (x) rotates continuously
**

with an angel of 2πk

The angle is measured counterclockwise.

If C is chosen to encircle a single isolated Equ. point X

The integer k is called the index of X

k = +1 for a node or focus, k = −1 for saddle.

**Inside any periodic orbit γ, there must be at least one equilibrium point.
**

Suppose the equilibrium points inside γ are hyperbolic, then if N is the

number of nodes and foci and S is the number of saddles, it must be that

N −S = 1.

**An equilibrium point is hyperbolic if the jacobian at that point has no
**

eigenvalue on the imaginary axis.

If the Equ point is not hyperbolic, then k may diﬀer from ±1

**It is useful in ruling out the existence of periodic orbits
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 52/53

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Index Theorem

**Example: The system
**

˙ x

1

= −x

1

+ x

1

x

2

˙ x

2

= x

1

+ x

2

−2x

1

x

2

has two equilibrium points at (0, 0) and (1, 1). The Jacobian:

_

∂f

∂x

_¸

¸

¸

¸

(0,0)

=

_

−1 0

1 1

_

;

_

∂f

∂x

_¸

¸

¸

¸

(1,1)

=

_

0 1

−1 −1

_

(0, 0) is a saddle point and (1, 1) is a stable focus.

**The only combination of Equ. points that can be encircled by a periodic
**

orbit is a single focus.

**Periodic orbit in other region such as that encircling both Equ. points are
**

ruled out.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 53/53

Introduction

Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

**Phase Plane Analysis: is a graphical method for studying second-order systems by
**

providing motion trajectories corresponding to various initial conditions. then examine the qualitative features of the trajectories. ﬁnally obtaining information regarding the stability and other motion patterns of the system.

It was introduced by mathematicians such as Henri Poincare in 19th century.

**http://en.wikipedia.org/wiki/Henri_Poincar%C3%A9
**

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 2/53

Introduction

Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Motivations

Importance of Knowing Phase Plane Analysis:

Since it is on second-order, the solution trajectories can be represented by carves in plane provides easy visualization of the system qualitative behavior. 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.

Disadvantage of Phase Plane Method: It is restricted to at most second-order and graphical study of higher-order is computationally and geometrically complex.

Dr. Farzaneh Abdollahi

Nonlinear Control

Lecture 2

3/53

Introduction

Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

Introduction Motivation Phase Plane Vector Field Diagram Isocline Method Qualitative Case 1: Case 2: Case 3: Case 4: Behavior of Linear Systems λ1 = λ2 = 0 Complex Eigenvalues, λ1,2 = α ± jβ Nonzero Multiple Eigenvalues λ1 = λ2 = λ = 0 Zero Eigenvalues

Local Behavior of Nonlinear Systems Limit Cycle

Dr. Farzaneh Abdollahi

Nonlinear Control

Lecture 2

4/53

Introduction

Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems

**Concept of Phase Plane
**

Phase plane method is applied to autonomous 2nd order system described as follows: x1 = f1 (x1 , x2 ) ˙ x2 = f2 (x1 , x2 ) ˙ f1 , f2 : R2 → R. System response (x(t) = (x1 (t), x2 (t))) to initial condition x0 = (x10 , x20 ) is a mapping from R to R2 . The x1 − x2 plane is called State plane or Phase plane The locus in the x1 − x2 plane of the solution x(t) for all t ≥ 0 is a curve named trajectory or orbit that passes through the point x0 The family of phase plane trajectories corresponding to various initial conditions is called Phase protrait of the system.

Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 5/53

(1) (2)

etc Vector Field Diagram: Revisiting (1) and (2): f1 (x) x = f (x) = ˙ . x = (x1 . 2) Dr. 1) + (2. 1) = (3. x2 ) ˙ ˙ ˙ f2 (x) To each vector (x1 . for x = (1. a corresponding vector (f1 (x1 .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems How to Construct Phase Plane Trajectories? Despite of exiting several routines to generate the phase portraits by computer. 2 Example: If f (x) = (2x1 . Some methods named: Isocline. x2 ). Farzaneh Abdollahi Nonlinear Control Lecture 2 6/53 . x2 ). x2 ). 1). it is useful to learn roughly sketch the portraits or quickly verify the computer outputs. delta method. x2 )) known as a vector ﬁeld is associated. Pell’s method. next point is (1. f2 (x1 . Vector ﬁeld diagram.

Example: Pendulum without friction x˙1 = x2 x˙2 = −10 sin x1 Nonlinear Control Lecture 2 7/53 Dr. the trajectory can be constructed. a vector ﬁeld diagram is obtained. ∴ starting from x0 and by using the vector ﬁeld with suﬃcient points. Farzaneh Abdollahi . Noting that dx2 = f2 vector ﬁeld at dx1 f1 a point is tangent to trajectory through that point.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Vector Field Diagram By repeating this for suﬃcient point in the state space.

x2 ) have the same tangent slope α. x2 ) An isocline with slope α is deﬁned as S(x) = α ∴ all the points on the curve f2 (x1 . x2 ) = αf1 (x1 . The responses x1 (t) and x2 (t) Only qualitative behavior can be concluded.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Isocline Method The term isocline derives from the Greek words for ”same slope. the slope of the trajectory at point x: S(x) = f2 (x1 . x2 ) dx2 = dx1 f1 (x1 . Dr. Note that the ”time” is eliminated here cannot be obtained directly. such as stable or oscillatory response.” Consider again Eqs (1) and (2). Farzaneh Abdollahi Nonlinear Control Lecture 2 8/53 .

Dr. Plot the curve S(x) = α in state-space (phase plane) 2. Draw small line with slope α.t. Repeat the process for suﬃcient number of α s.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Isocline Method The algorithm of constructing the phase portrait by isocline method: 1. Farzaneh Abdollahi Nonlinear Control Lecture 2 9/53 . Note that the direction of the line depends on the sign of f1 and f2 at that point. the phase plane is full of isoclines. 3.

Dr.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Example: Pendulum without Friction Consider the dynamics x1 = x2 . Farzaneh Abdollahi Nonlinear Control Lecture 2 10/53 . 2 The closed curve trajectory conﬁrms marginal stability of the system. conditions can be obtained by using the given algorithm The response for x0 = ( π . 0) is depicted in Fig. x2 = −sinx1 ∴S(x) = ˙ ˙ Isoclines: x2 = −1 c sinx1 −sinx1 x2 =c Trajectories for diﬀerent init.

x2 = −0. 0) 2 The trajectory shrinks like an spiral converging to the origin Dr.5+c sinx1 −0.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Example: Pendulum with Friction Dynamics of pendulum with friction: x1 = x2 . Farzaneh Abdollahi Nonlinear Control Lecture 2 11/53 .5−sinx1 x2 =c Similar Isoclines but with diﬀerent slopes Trajectory is drawn for x0 = ( π .5x2 − sinx1 ∴S(x) = ˙ ˙ Isoclines: x2 = −1 0.

Jr has one of the following forms: λi : real & distinct λi : real & multiple λi : complex Jr Jr Jr = = = λ1 0 0 λ2 λ k 0 λ α −β β α 12/53 .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Qualitative Behavior of Linear Systems First we analyze the phase plane of linear systems since the behavior of nonlinear systems around equilibrium points is similar of linear ones For LTI system: x = Ax. Farzaneh Abdollahi The system behavior is diﬀerent at each Lecture 2 case Nonlinear Control . x0 : initial state x(t) = Me Jr t M −1 x0 ˙ Jr : Jordan block of A. k = 0. M : Matrix of eigenvectors M −1 AM = Jr Depending on the eigenvalues of A. A ∈ R2×2 . Dr. 1.

z02 ): z1 (t) = z10 e λ1 t .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 1: λ1 = λ2 = 0 In this case M = [v1 v2 ] where v1 and v2 are real eigenvectors associated with λ1 and λ2 To transform the system into two decoupled ﬁrst-order diﬀ equations. Dr. C = z20 /(z10 )λ2 /λ1 (3) Phase portrait is obtained by changing C ∈ R and plotting (3). z2 (t) = z20 e λ2 t eliminating t z2 = Cz1 2 λ /λ1 . The phase portrait depends on the sign of λ1 and λ2 . Farzaneh Abdollahi Nonlinear Control Lecture 2 13/53 . let z = M −1 x: z1 = λ1 z1 ˙ z2 = λ2 z2 ˙ The solution for initial states (z01 .

as they approach to origin parallel to z2 axis.1: λ2 < λ1 < 0 t → ∞ ⇒ the terms e λ1 t and e λ2 t tend to zero Trajectories from entire state-space tend to origin the equilibrium point x = 0 is stable node. Farzaneh Abdollahi . as they are far from origin. so slope is zero as z1 −→ 0 inﬁnity as z1 −→ ∞. Slope of the curves: λ2 < λ1 < 0 dz2 dz1 = C λ2 z 1 λ1 (λ2 /λ1 −1) λ2 /λ1 > 1. ∴ The trajectories are tangent to z1 axis. e λ2 t → 0 faster λ2 is fast eignevalue and v2 is fast eigenvector. Nonlinear Control Lecture 2 14/53 Dr.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 1.

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 1. trajectories are sliding along z1 axis In X plane also trajectories are: tangent to the slow eigenvector v1 for near origin parallel to the fast eigenvector v2 for far from origin Dr.1: λ2 < λ1 < 0 Since z2 approaches to zero faster than z1 . Farzaneh Abdollahi Nonlinear Control Lecture 2 15/53 .

2: λ2 > λ1 > 0 t → ∞ ⇒ the terms e λ1 t and e λ2 t grow exponentially. so The shape of the trajectories are the same. Farzaneh Abdollahi Nonlinear Control Lecture 2 16/53 .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 1. with opposite directions The equilibrium point is socalled unstable node Dr.

but increasing in z1 direction tangent to z1 as |z1 | → ∞ and tangent to z2 as |z1 | → 0 Dr. decreasing in z2 direction.3: λ2 < 0 < λ1 t → ∞ ⇒ e λ2 t −→ 0.so λ2 : stable eigenvalue. but e λ1 t −→ ∞.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 1. v1 : unstable eigenvector Trajectories are negative exponentials since Trajectories are λ2 λ1 is negative. Farzaneh Abdollahi Nonlinear Control Lecture 2 17/53 . v2 : stable eigenvector λ1 : unstable eigenvalue.

3: λ2 < 0 < λ1 The exceptions of this hyperbolic shape: two trajectories along z2 -axis → 0 as t → 0.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 1. Dr. stable trajectories are along v2 . Farzaneh Abdollahi Nonlinear Control Lecture 2 18/53 . called unstable trajectories This equilibrium point is called saddle point Similarly in X plane. called stable trajectories two trajectories along z1 -axis → ∞ as t → 0. but unstable trajectories are along the v1 For λ1 < 0 < λ2 the direction of the trajectories are changed.

The spiral converges to origin =⇒ Stable Focus. α > 0: as t → ∞ r → ∞ and angle θ is rotating.2 = α ± jβ z˙1 = αz1 − βz2 z˙2 = βz1 + αz2 The solution is oscillatory =⇒ polar coordinates (r = z2 2 2 z1 + z2 . α = 0: Trajectories are circles with radius r0 =⇒ Center Dr. θ = tan−1 ( z1 )) r = αr r (t) = r0 e αt ˙ ˙ θ = β θ(t) = θ0 + βt This results in Z plane is a logarithmic spiral where α determines the form of the trajectories: α < 0 : as t → ∞ r → 0 and angle θ is rotating. Farzaneh Abdollahi Nonlinear Control Lecture 2 19/53 . The spiral diverges away from origin =⇒ Unstable Focus.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 2: Complex Eigenvalues. λ1.

Farzaneh Abdollahi Nonlinear Control Lecture 2 20/53 .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 2: Complex Eigenvalues.2 = α ± jβ Dr. λ1.

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 3: Nonzero Multiple Eigenvalues λ1 = λ2 = λ = 0 Let z = M −1 x: z1 = λz1 + kz2 . When the eignevectors are diﬀerent k = 0: similar to Case 1. Farzaneh Abdollahi Nonlinear Control Lecture 2 21/53 . for λ < 0 is stable. Decaying rate is the same for both modes (λ1 = λ2 ) trajectories are lines Dr. z2 (t) = z20 e λt z2 z10 k + ln z1 = z2 z20 λ z20 Phase portrait are depicted for k = 0 and k = 1. λ > 0 is unstable. ˙ λt z2 = λz2 ˙ the solution is :z1 (t) = e (z10 + kz20 t).

k = 1 is more complex.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 3: Nonzero Multiple Eigenvalues λ1 = λ2 = λ = 0 There is no fast-slow asymptote. but it is still similar to Case 1: Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 22/53 .

v1 ∈ N (A). z1 = 0. z2 (t) = z20 e λ2 t Phase portrait depends on sign of λ2 : λ2 < 0: Trajectories converge to equilibrium line λ2 > 0: Trajectories diverge from equilibrium line Nonlinear Control Lecture 2 23/53 Dr. Farzaneh Abdollahi . v2 : corresponding eigenvectors.1: One eigenvalue is zero λ1 = 0. v1 .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 4. z2 = λ2 z2 ˙ ˙ solution: z1 (t) = z10 . λ2 = 0 A is singular in this case Every vector in null space of A is an equilibrium point There is a line (subspace) of equilibrium points M = [v1 v2 ] .

Farzaneh Abdollahi Nonlinear Control Lecture 2 24/53 .1 and 4. z2 (t) = z20 z1 linearly increases/decreases base on the sign of z20 z2 axis is equilibrium subspace in Z-plane Dotted line is equilibrium subspace The diﬀerence between Case 4. Dr.2: all trajectories start oﬀ the equilibrium set move parallel to it.2: Both eigenvalues zero λ1 = λ2 = 0 Let z = M −1 x z1 = z2 . z2 = 0 ˙ ˙ solution: z1 (t) = z10 + z20 t.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Case 4.

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems As Summary: Six types of equilibrium points can be identiﬁed: stable/unstable node saddle point stable/ unstable focus center Type of equilibrium point depends on sign of the eigenvalues If real part of eignevalues are Positive If real part of eignevalues are Negative unstability stability All properties for linear systems hold globally Properties for nonlinear systems only hold locally Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 25/53 .

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Local Behavior of Nonlinear Systems Qualitative behavior of nonlinear systems is obtained locally by linearization around the equilibrium points Type of the perturbations and reaction of the system to them determines the degree of validity of this analysis A simple example: Consider the linear perturbation case A −→ A + ∆A. For eigenvalues on the jω axis no matter how small perturbation is. it changes the sign of eigenvalue. Farzaneh Abdollahi Nonlinear Control Lecture 2 26/53 . Therefore node or saddle point or focus equilibrium point remains the same under small perturbations This analysis is not valid for a center equilibrium point Dr. where ∆A ∈ R2×2 : small perturbation Eigenvalues of a matrix continuously depend on its parameters Positive (Negative) eigenvalues of A remain positive (negative) under small perturbations.

f2 are continuously diﬀerentiable about (x10 . Farzaneh Abdollahi . x20 ) Since we are interested in trajectories near (x10 . nonlinear systems can have multiple isolated equilibria. x20 ) are equilibrium points of x1 ˙ x2 ˙ = = f1 (x1 .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Multiple Equilibria Linear systems can have an isolated equilibrium point or a continuum of equilibrium points (When detA = 0) Unlike linear systems. x2 ) f2 (x1 . x2 = y2 + x20 y1 . x20 ). x2 ) (4) f1 . y2 are small perturbations form equilibrium point. Qualitative behavior of second-order nonlinear system can be investigated by generating phase portrait of system globally by computer programs linearize the system around equilibria and study the system behavior near them without drawing the phase portrait Let (x10 . Nonlinear Control Lecture 2 27/53 Dr. deﬁne x1 = y1 + x10 .

T . Farzaneh Abdollahi a11 a21 a12 a22 = ∂f1 ∂x1 ∂f2 ∂x1 Lecture 2 ∂f1 ∂x2 ∂f2 ∂x2 = ∂f ∂x x0 28/53 Nonlinear Control .O. are negligible ∂fi y1 = a11 y1 + a12 y2 ˙ . A = ˙ Dr. x20 ) + ˙ ˙ ˙ 0 ∂f1 ∂x1 ∂f2 ∂x1 y1 + (x10 .O. aij = .O.T . 2 y2 = a21 y1 + a22 y2 ˙ ∂x x0 The equilibrium point of the linear system is (y1 = y2 = 0) y = Ay .x20 ) x2 = x20 + y2 = f2 (x10 . (x10 . H. x20 ) + ˙ ˙ ˙ 0 y1 + (x10 .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Qualitative Behavior Near Equilibrium Points Expanding (4) into its Taylor series x1 = x10 + y1 = f1 (x10 . (x10 .x20 ) ∂f1 ∂x2 ∂f2 ∂x2 y2 + H.x20 ) For suﬃciently small neighborhood of equilibrium points.x20 ) y2 + H.T. i = 1.

or a stable (unstable) focus or a saddle point. then in a small neighborhood of the equilibrium point. the trajectory of the nonlinear system will behave like a stable (unstable) node. Farzaneh Abdollahi Nonlinear Control Lecture 2 29/53 . Dr. or a stable (unstable) focus or a saddle point.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Qualitative Behavior Near Equilibrium Points Matrix ∂f ∂x is called Jacobian Matrix. The trajectories of the nonlinear system in a small neighborhood of an equilibrium point are close to the trajectories of its linearization about that point: if the origin of the linearized state equation is a stable (unstable) node.

2v .79x1 + 229.285.063.31x1 + 83. 0.61).884. Q2 = (0.76x1 − 103.5K Ω. 0.5x2 + 1. Farzaneh Abdollahi Nonlinear Control Lecture 2 30/53 .21) Dr.62x1 − 226.72x1 equilibrium points (x1 = x2 = 0): ˙ ˙ Q1 = (0.5[−h(x1 ) + x2 ] ˙ x2 = 0. R = 1. current in mA x1 = 0.758).2[−x1 − 1. C = 2pF . 0. time in nanosecond. L = 5µH. Q3 = (0.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Example: Tunnel Diode Circuit x1 = ˙ x2 = ˙ 1 [−h(x1 ) + x2 ] C 1 [−x1 − Rx2 + u] L u = 1.2] ˙ 2 3 4 5 Suppose h(x1 ) = 17.

all trajectories approach either Q1 or Q3 . Except for two special trajectories which approach Q2 . Near equilibrium points Q1 and Q3 are stable nodes.Q2 is like saddle point. Farzaneh Abdollahi .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Example: Tunnel Diode Circuit The global phase portrait is generated by a computer program is shown in Fig. The two special trajectories from a curve that divides the plane into two halves with diﬀerent behavior (separatrix curves). All trajectories originating from left side of the curve approach to Q1 All trajectories originating from left side of the curve approach to Q3 Nonlinear Control Lecture 2 31/53 Dr.

0. Q3 : −3.76 − 207. A2 = Q3 = (0.25 saddle point .4 stable node Q1 = (0.33 stable node .24x1 + 418.3 0. Q2 . λ1 = −1. λ2 = −0. Farzaneh Abdollahi ∴ similar results given from global phase portrait.758).3 .5 −0.2 0.598 −0. Lecture 2 32/53 .285.2 −0.6x1 dx1 ´ h(x1 ) = Evaluate the Jacobian matrix at the equilibriums Q1 .063.427 0. λ2 = −0.82 −0. A1 = Q2 = (0.33.2 1. λ1 = −3.5 −0. λ1 = 1. 0.86x1 − 905.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Tunnel Diode:Qualitative Behavior Near Equilibrium Points Jacobian matrix ∂f ∂x = ´ 0.57.61). λ2 = −0. 0.2 −0.5 −0.77. A3 = −1.884.58x1 + 688.5h(x1 ) 0.5 −0.3 dh 2 3 4 = 17.21).3 Nonlinear Control Dr.

Triggering formQ1 to Q3 or vice versa is achieved by changing the load line Dr.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Tunnel Diode Circuit In practice. Farzaneh Abdollahi Nonlinear Control Lecture 2 33/53 . since it has two steady-state operating points. There are only two stable equilibrium points: Q1 or Q3 . Equilibrium point at Q2 in never observed. Even if set up the exact initial conditions corresponding t Q2 . the ever-present physical noise causes the trajectory to diverge from Q2 Such circuit is called bistable.

x2 = r sin θ ⇒ r = −µr 3 . Example: 2 2 x1 = −x2 − µx1 (x1 + x2 ) ˙ 2 2 x2 = x1 − µx2 (x1 + x2 ) ˙ It has equilibrium point at origin x ∗ = 0. then the qualitative behavior of nonlinear system near the equilibrium point could be quite distinct from the linearized one. A= 0 −1 1 0 ⇒ λ1. Nonlinear Control Lecture 2 34/53 Dr.2 = ±j ⇒ center Now consider the nonlinear system ˙ x1 = r cos θ.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Special case: If the Jacobian matrix has eigenvalues on jω. θ = 1 ˙ ∴ nonlinear system is stable if µ > 0 and is unstable if µ < 0 ∴ the qualitative behavior of nonlinear and linearized one are diﬀerent. Farzaneh Abdollahi .

θ0 = tan−1 z10 Such oscillation where there is a continuum of closed orbits is referred to harmonic oscillator. Farzaneh Abdollahi . calling periodic orbit or closed orbit. z2 = r0 sin(βt + θ0 ) z20 2 2 r0 = z10 + z20 .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Limit Cycle A system oscillates when it has a nontrivial periodic solution x(t + T ) = x(t). for someT > 0 The word ”nontrivial” is used to exclude the constant solutions. The image of a periodic solution in the phase portrait is a closed trajectory. The origin of the system is a center. ∀t ≥ 0. Nonlinear Control Lecture 2 35/53 r0 : amplitude of oscillation Dr. We have already seen oscillation of linear system with eigenvalues ±jβ. and the trajectories are closed the solution in Jordan form: z1 (t) = r0 cos(βt + θ0 ).

A practical nonlinear oscillator can be build such that The nonlinear oscillator is structurally stable The amplitude of oscillation (at steady state) is independent of initial conditions.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Limit Cycle The physical mechanism leading to these oscillations is a periodic exchange of energy stored in the capacitor (electric ﬁeld) and the inductor (magnetic ﬁeld). These problems can be eliminated in nonlinear oscillators. The linear oscillator is not structurally stable The amplitude of the oscillation depends on the initial conditions. any small perturbations Dr. We have seen that such oscillation is not robust destroy the oscillation. Farzaneh Abdollahi Nonlinear Control Lecture 2 36/53 .

a limit cycle is deﬁned as an isolated closed orbit. they are not isolated. Farzaneh Abdollahi . closed: indicating the periodic nature of the motion 2. Depends on trajectories motion pattern in vicinity of limit cycles. there are three type of limit cycle: Stable Limit Cycles: as t → ∞ all trajectories in the vicinity converge to the limit cycle. For limit cycle the trajectory should be 1. The mass spring damper does not have limit cycle. isolated: indicating limiting nature of the cycle with nearby trajectories converging to/ diverging from it. Unstable Limit Cycles: as t → ∞ all trajectories in the vicinity diverge from the limit cycle.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Limit Cycle On phase plane. Nonlinear Control Lecture 2 37/53 Dr. Semi-stable Limit Cycles: as t → ∞ some trajectories in the vicinity converge to/ and some diverge from the limit cycle.

Farzaneh Abdollahi Nonlinear Control Lecture 2 38/53 .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Limit Cycle Dr.

x2 =: rsin(θ)) r = −r (r 2 − 1) ˙ ˙ θ = −1 2 2 If trajectories start on the unit circle (x1 (0) + x2 (0) = r 2 = 1). Nonlinear Control Lecture 2 39/53 Dr. r < 1 =⇒ r > 0 =⇒ trajectories converges to the unit circle from inside.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Example1. Unit circle is a stable limit cycle for this system.a: stable limit cycle 2 2 x1 = x2 − x1 (x1 + x2 − 1) ˙ 2 2 x2 = −x1 − x2 (x1 + x2 − 1) ˙ Polar coordinates (x1 := rcos(θ). ˙ r > 1 =⇒ r < 0 =⇒ trajectories converges to the unit circle from ˙ outside. Farzaneh Abdollahi . then r = 0 =⇒ The trajectory will circle the origin of the phase plane with ˙ 1 period of 2π .

b: unstable limit cycle 2 2 x1 = x2 + x1 (x1 + x2 − 1) ˙ 2 2 x2 = −x1 + x2 (x1 + x2 − 1) ˙ Polar coordinates (x1 := rcos(θ). then r = 0 =⇒ The trajectory will circle the origin of the phase plane with ˙ 1 period of 2π .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Example1. r > 1 =⇒ r > 0 =⇒ trajectories diverges from the unit circle from ˙ outside. Nonlinear Control Lecture 2 40/53 Dr. x2 =: rsin(θ)) r = r (r 2 − 1) ˙ ˙ θ = −1 2 2 If trajectories start on the unit circle (x1 (0) + x2 (0) = r 2 = 1). Unit circle is an unstable limit cycle for this system. Farzaneh Abdollahi . r < 1 =⇒ r < 0 =⇒ trajectories diverges from the unit circle from ˙ inside.

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Example1. then r = 0 =⇒ The trajectory will circle the origin of the phase plane with ˙ 1 period of 2π . r < 1 =⇒ r < 0 =⇒ trajectories diverges from the unit circle from ˙ inside. Farzaneh Abdollahi .c: semi stable limit cycle 2 2 x1 = x2 − x1 (x1 + x2 − 1)2 ˙ 2 2 x2 = −x1 − x2 (x1 + x2 − 1)2 ˙ Polar coordinates (x1 := rcos(θ). Unit circle is a semi-stable limit cycle for this system. Nonlinear Control Lecture 2 41/53 Dr. r > 1 =⇒ r < 0 =⇒ trajectories converges to the unit circle from ˙ outside. x2 =: rsin(θ)) r = −r (r 2 − 1)2 ˙ ˙ θ = −1 2 2 If trajectories start on the unit circle (x1 (0) + x2 (0) = r 2 = 1).

The boundary of the set is connected if for connecting any two points belong to boundary of the set there is a line which does not cross the set exmaple of connected but not simply connected set Nonlinear Control Lecture 2 42/53 Dr. If f is not identically zero over any 1 2 subregion of Ω and does not change sign in Ω. then Ω contain no limit cycle for the nonlinear system x1 = f1 (x1 . x2 ) ˙ x2 = f2 (x1 . Farzaneh Abdollahi . there is a line which remains in the set.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Bendixson’s Criterion: Nonexistence Theorem of Limit Cycle Gives a suﬃcient condition for nonexistence of a periodic solution: Suppose Ω is simply connected region in 2-dimention space in this region ∂f ∂f we deﬁne f = ∂x1 + ∂x2 . x2 ) ˙ Simply connected set: the boundary of the set is connected + the set is connected Connected set: for connecting any two points belong to the set.

ndl = S fds = 0 S is enclosed by L) S ∂f2 ∂f1 + ∂x1 ∂x2 dx1 dx2 = 0 This is true if f = 0 ∀x ∈ S or f changes sign in S ∂f ∂f This is in contradiction with the assumption that ∂x1 + ∂x2 does not vanish 1 2 and does not change sign there is no closed trajectory. Nonlinear Control Lecture 2 43/53 Dr. dx1 ∴ Along a closed curve L of a limit cycle: (f2 dx1 − f1 dx2 ) = 0 L Using Stoke’s Theorem: ( L f . Farzaneh Abdollahi .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Bendixson’s Criterion Proof by contradiction: f2 Recall that dx2 = f1 =⇒ f2 dx1 − f1 dx2 = 0.

Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Nonexistence Theorem of Periodic Solutions for Linear Systems Suﬃcient condition for nonexistence of a periodic solution in linear systems: x1 = a11 x1 + a12 x2 ˙ x2 = a21 x1 + a22 x2 ˙ ∂f ∴ ∂x1 + 1 ∂f2 ∂x2 = a11 + a22 = 0=⇒ no periodic sol. This is consistent with eigenvalue analysis form of center point which is obtained for periodic solutions: λ2 − (a11 + a22 ) λ + (a11 a22 − a12 a21 ) = 0 center ∴ a11 + a22 = 0. a11 a22 − a12 a21 > 0 Nonlinear Control Lecture 2 44/53 Dr. Farzaneh Abdollahi .

Dr.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Limit Cycle Example for nonexistence of limit cycle 2 x1 = g (x2 ) + 4x1 x2 ˙ 2 x2 = h(x1 ) + 4x1 x2 ˙ ∂f1 ∂x1 ∂f2 ∂x2 ∴ + 2 2 = 4(x1 + x2 ) > 0 ∀x ∈ R2 No limit cycle exist in R2 for this system. Note that: there is no equivalent theorem for higher order systems. Farzaneh Abdollahi Nonlinear Control Lecture 2 45/53 .

it must have a periodic solution. Hence. M contains no equilibrium point or contains only one equilibrium point such ∂f that the Jacobian matrix ∂x at this point has eigenvalues with positive real parts (unstable focus or node). Dr. 1. if the positive limit set contains no equilibrium point. Farzaneh Abdollahi Nonlinear Control Lecture 2 46/53 . 2. Every trajectory starting in M stays in M for all future time =⇒ M contains a periodic solution The idea behind the theorem is that all possible shape of limit points in a plane (R2 ) are either equilibrium points or periodic solutions.t.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Poincare-Bendixson Criterion: Existence Theorem of Limit Cycle If there exists a closed and bounded set M s.

A solution could be bounded in R3 . No equivalent theorem for Rn . in vicinity of that equilibrium point all trajectories move away By excluding the vicinity of unstable node/focus. Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 47/53 . n ≥ 3.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Poincare-Bendixson Criterion: Existence Theorem of Limit Cycle If M has unstable node/focus. but neither it is periodic nor it tends to a periodic solution. the set M is free of equilibrium and all trajectories are trapped in it.

Nonlinear Control Lecture 2 48/53 Dr.B. 1 > η2 > 0 ˙ The area found by the circles with radius 1 − η2 and 1 + η1 satisﬁes the condition of the P. Farzaneh Abdollahi . η1 > 0 ˙ r ≥ 0for r ≤ 1 − η2 .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Example for Existence Theorem of Limit Cycle 2 2 x1 = −x2 − x1 (x1 + x2 − 1) ˙ 2 2 x2 = x1 − x2 (x1 + x2 − 1) ˙ Polar coordinate: r = (1 − r 2 )r ˙ ˙ θ = −1 r ≤ 0 for r ≥ 1 + η1 . theorem a periodic solution exists.

c2 > 0. Nonlinear Control Lecture 2 49/53 Dr.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Existence Theorem of Limit Cycle A method to investigate whether or not trajectories remain inside M: Consider a simple closed curve V (x) = c. for some c > 0. V (x) ≤ 0 on the boundary of M. V (x) > 0 tangent to the curve if f (x). Farzaneh Abdollahi . For a set of the form M = {V (x) ≤ c}. W (x) ≥ 0 on W (x) = c1 . trajectories trapped inside M if f (x). V (x) = ∂V ∂V f1 (x) + f2 (x) < 0 ∂x1 ∂x2 V (x) is negative: outward if f (x). trajectories remain inside M if f (x). where V (x) is a continuously diﬀerentiable function The vector f at a point x on the curve points inward if the inner product of f and the gradient vector f (x). For an annular region of the form M = {W (x) ≥ c1 and V (x) ≤ c2 } for some c1 . V (x) = 0. Trajectories can leave a set only if the vector ﬁled points outward at some points on the boundary. V (x) ≤ 0 on V (x) = c2 and f (x).

The Jacobian matrix ∂f ∂x = x=0 2 2 1 − 3x1 − x2 1 − 2x1 x2 2 2 −2 − 2x1 x2 1 − x1 − 3x2 = x=0 1 1 −2 1 has eigenvalues at 1 ± j 2 2 2 Let M = V (x) ≤ c. √ M is bounded and contains eigenvalues with positive real parts Nonlinear Control Lecture 2 50/53 Dr.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Example for Existence Theorem of Limit Cycle Consider the system 2 2 x1 = x1 + x2 − x1 (x1 + x2 ) ˙ 2 2 x2 = −2x1 + x2 − x2 (x1 + x2 ) ˙ The system has a unique equilibrium point at the origin. Farzaneh Abdollahi . where V (x) = x1 + x2 .

∴ by PB criterion.5 ensures that all trajectories trapped inside M. Dr. Farzaneh Abdollahi Nonlinear Control Lecture 2 51/53 . there exits at least one periodic orbit.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Example for Existence Theorem of Limit Cycle On the surface V (x) = c. we have ∂V ∂V 2 2 f1 (x) + f2 (x) = 2x1 [x1 + x2 − x1 (x1 + x2 )] ∂x1 ∂x2 2 2 +2x2 [−2x1 + x2 − x2 (x1 + x2 )] 2 2 2 2 = 2(x1 + x2 ) − 2(x1 + x2 )2 − 2x1 x2 2 2 2 2 2 2 ≤ 2(x1 + x2 ) − 2(x1 + x2 )2 + (x1 + x2 ) = 3c − 2c 2 Choosing c > 1.

it must be that N − S = 1. then k may diﬀer from ±1 It is useful in ruling out the existence of periodic orbits Nonlinear Control Lecture 2 52/53 Dr. An equilibrium point is hyperbolic if the jacobian at that point has no eigenvalue on the imaginary axis. Farzaneh Abdollahi .Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Index Theorem Let C be a simple closed curve not passing through any equilibrium point Consider the orientation of the vector ﬁeld f (x) at a point p ∈ C . If the Equ point is not hyperbolic. k = −1 for saddle. there must be at least one equilibrium point. If C is chosen to encircle a single isolated Equ. Suppose the equilibrium points inside γ are hyperbolic. point X The integer k is called the index of X k = +1 for a node or focus. Inside any periodic orbit γ. then if N is the number of nodes and foci and S is the number of saddles. Let p traverse C counterclockwise the vector f (x) rotates continuously with an angel of 2πk The angle is measured counterclockwise.

Farzaneh Abdollahi . Nonlinear Control Lecture 2 53/53 Dr.0) −1 0 1 1 . points are ruled out. 0) and (1. 1). 1) is a stable focus. 0) is a saddle point and (1. Periodic orbit in other region such as that encircling both Equ.1) 0 1 −1 −1 (0. points that can be encircled by a periodic orbit is a single focus. The Jacobian: ∂f ∂x = (0. ∂f ∂x = (1.Introduction Phase Plane Qualitative Behavior of Linear Systems Local Behavior of Nonlinear Systems Index Theorem Example: The system x1 = −x1 + x1 x2 ˙ x2 = x1 + x2 − 2x1 x2 ˙ has two equilibrium points at (0. The only combination of Equ.

Hybrid Control to Approach Chaos Synchronization of Uncertain DUFFING Oscillator Systems with External Disturbance

by International Journal of Robotics, Theory and Applications

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd