Scribd
Upload
13 views

NonLinear Lecture 1.3

1) Equilibrium points, also called critical points or singular points, occur when the derivatives of the state variables are equal to zero. 2) For a 2nd order linear system represented in state space form, the equilibrium points must lie on the x1-axis. 3) The solutions for both state variables are of the same form, determined by the characteristic equation associated with the eigenvalues of the system matrix. 4) The nature of the equilibrium point (node, saddle, focus, center) and its stability are determined by the eigenvalues.

Uploaded by

jiales225
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
13 views

NonLinear Lecture 1.3

1) Equilibrium points, also called critical points or singular points, occur when the derivatives of the state variables are equal to zero. 2) For a 2nd order linear system represented in state space form, the equilibrium points must lie on the x1-axis. 3) The solutions for both state variables are of the same form, determined by the characteristic equation associated with the eigenvalues of the system matrix. 4) The nature of the equilibrium point (node, saddle, focus, center) and its stability are determined by the eigenvalues.

Uploaded by

jiales225
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 16

Non-Linear Systems and the Phase

Plane
Lecture 1.3

Paul White
Outline
• “Quotes” and preamble.
• State-space
• Phase plots
• Trajectories and phase portraits
• Equilibrium points
• Solutions of linear equations

2
Critical Points/ Equilibrium Points/
Singular Points
• Points where f= 1 ( x1 , x2 ) 2 ( x1 , x2 )
f= 0 are called singularities,
critical points or equilibrium points.
=
• Clearly at these points: x1 0= and x2 0 so x1 and x2 are not
changing – they have reached an equilibrium.

• An autonomous linear system of equations has the form


x = Ax
• For this system there is generally a single equilibrium point at
the origin x=0 (there are only degenerate counter examples).
• For a non-linear system there may be multiple equilibrium
points which can be away from the origin.
3
Equilibrium Points for 2nd Order ODEs

• For a state-space representation derived from a 2nd order ODE,


i.e. a representation of the form
x1 = x2
x2 = f 2 ( x1 , x2 )

• The equilibrium points are given by


x2 = 0
f 2 ( x1 ,0 ) = 0

i.e. the equilibrium points must lie on the x1-axis.


• Note this not true for general state-space models.
4
Forms of Solution for 1st Order Linear
Systems
• Initially let us review the form of solutions for linear first
order equations.
x =
λx Ceλt
⇒ x=

• The form of the solution depends on the value of λ.


– If λ is real and negative, x decays exponentially (stable).
– If λ is real and positive, x grows exponentially (unstable).
– If λ is zero, the solution is a constant (degenerate solution)
– If λ is complex
• If Re{λ} is negative, solution oscillates as it decays (stable).
• If Re{λ} is positive, solution oscillates as it grows (unstable).

5
Solutions for 2nd Order Linear System

• A general pair of 1st order linear differential equations:


x1 ax1 + bx2
= (1)
=x Ax ⇒
2 cx1 + dx2
x= (2)

where a, b, c and d are the elements of A.


• There are several methods for solving these equations, we
shall adopt a very direct method.
• Multiplying (2) by b and differentiating (1) wrt to t gives

x1 ax1 + bx2
= (3)
=
bx2 bcx1 + bdx2 (4)

6
• Using (2) to express bx2 and substituting into (4) gives
bx2 =bcx1 + d ( x1 − ax1 ) (5)
• Finally using (3) to express bx2 and substitute into (5) gives

x1 − ax1 = bcx1 + d ( x1 − ax1 )



x1 − ( a + d ) x1 + ( ad − bc ) x1 =
⇒  0

x1 − αx1 + βx1 = 0
where α = a + d which is the trace of A and β = ad − bc which
is the determinant of the matrix A.
• Similarly, for x2 one gets an equation of the same form:
x2 − ( a + d ) x2 + ( ad − bc ) x2 =
 0

x2 − αx2 + βx2 = 0

• So there is a single form of equation, 


x − αx + βx = 0, which
controls x1 and x2. 7
Form of Solutions
• The solution to 
x − αx + βx = 0

• Has the form x ( t ) C1eλ t + C2 eλ t


= 1 2

where C1 and C2 are arbitrary constants dependent upon the


initial conditions, and λ1 and λ2 are the solutions to the
characteristic equation:
λ 2 − αλ + β = 0
α + α 2 − 4β α − α 2 − 4β
= λ1 = λ2
2 2
λ1 and λ2 are the eigenvalues of the system matrix A
As in x = Ax
8
Eigenvalues of A
• To show λ1 and λ2 are the eigenvalues of A.
a b 
A= 
 c d 
• Eigenvalues of A are given by solving
a − λ b 
= A − λI =  0
 c d − λ
( a − λ )( d − λ ) − bc
= 0 ⇒ λ 2 − ( a + d ) λ + ad − bc
= 0
⇒ λ 2 − αλ + β = 0
• Using definitions of α and β from previous slides.
• This is exactly the equation from the last slide, demonstrating
that λ1 and λ2 are the eigenvalues of A.
9
Possible Forms of Solution
• λ1 and λ2 are real and both positive
– Unstable solution
• λ1 and λ2 are real and both negative
– Stable solution
• λ1 and λ2 are real and have opposite signs
– Unstable solution
• λ1 and λ2 are complex conjugates and positive real parts
– Unstable solution
• λ1 and λ2 are complex conjugates and negative real parts
– Stable solution
• λ1 and λ2 are complex conjugates and zero real parts
– Marginally stable solution

10
Forms of Solutions in Phase Plane (1)

• Node (real λ1 and λ2 of same sign)


– Unstable node, λ1 and λ2 positive
– Stable node, λ1 and λ2 negative

Stable Node Unstable Node

11
Forms of Solutions in Phase Plane (2)

• Saddle Point (λ1 and λ2 opposite signs)

12
Forms of Solutions in Phase Plane (3)

• Focus (complex λ1 and λ2 assuming a, b, c and d are real, then


λ1 = λ2 * )
– Unstable (positive real λ1 and λ2)
– Stable (negative real λ1 and λ2)
Stable Unstable

13
Forms of Solutions in Phase Plane (4)

• Centre (zero real parts of λ1 and λ2)

14
Comments
• All of the forms of solutions around an equilibrium point in
the phase plane are the result of combining 2 first order
solutions:
– A stable node is 2 stable non-oscillating first order equations
– An unstable node is 2 unstable first order equations
– A stable focus is 2 stable oscillating first order equations
– An unstable focus is 2 unstable oscillating first order equations
– A centre is 2 non-decaying first order equations
• These structures are typically rotated in the phase plane.
• The above is most readily apparent when the solution is
expressed through matrix exponentials (not as shown here).

15
Information in the Eigenvectors
• The eigenvalues of A allow one to determine the nature of an
equilibrium point (i.e. whether it is a focus, node, saddle or
centre and whether it is stable or unstable).
• The eigenvectors define the directions along which trajectories
converge.
• For example, the eigenvector associated with a positive
eigenvalue, defines the line trajectories approach as time tends
to infinity.

16

You might also like