Académique Documents
Professionnel Documents
Culture Documents
x 1 x 2
1 5
x 2 x1 x1 x 2
16
Calculate the equilibrium points for the system.
Plot the phase portrait of the system.
Solution:
The equilibrium points must be stationary. Therefore for the first system we
have
0 x2 0 x2
1 5
0 x1 x1 x 2
1 5
0 x1 x1 x1 1 0.0625x14
16
16
0 x2
1 5
0 x1 x1 x1 1 0.0625x14
16
roots([-1/16 0 0 0 1])
x1=0
ans =
-2.0000
-0.0000 + 2.0000i
-0.0000 - 2.0000i
2.0000
xe=[(0,0),(2,0),(-2,0)]
f1 f1
x x 2 0 1
J 1 5 4
f 2 f 2 1 x1 1
16
x1
x 2
0 1 0 1
0 1
J xe1( 0,0) J xe 2( 2, 0)
1
5
( 2) 4
1 4 1
1 1 16
eig (J) eig (J )
The same result is
obtained for xe3 (2,0)
- 0.5000 + 0.8660i 1.5616
- 0.5000 - 0.8660i - 2.5616 Saddle points
x1dot = x2;
2
0
x
x2dot = -x1+(1/16)*x1.^5-x2;
-0.5
quiver(x1,x2,x1dot,x2dot)
xlabel('x_1') -1
ylabel('x_2') -1.5
-2
-3 -2 -1 0 1 2 3
x1
Example 2. Show that the origin of the system is stable, using a suitable
Lyapunov function.
x 1 x 2
x 2 x13 x 32
1 4 1 2
V ( x ) x1 x 2
4 2
x2
V( x ) x1 , x 2 3
3
3
( x ) dV V dx V f ( x )
V
1x x 2
dt x dt V 1 2 2
(x) x 3x x x 3 x 3
1 2
V V f1 ( x ) (x) x 4 0
, f ( x ) V 2
1x x 2 2
The system is stable in the sense of Lyapunov.
Example 3:
R(s) + 1 y 3 C(s)
s2 1 s 1
-
y3
s N
Find the describing function of the nonlinear element N of the control system.
1
sin t
0.6 ODD FUNCT ION
3 0.4
4
0.2
w y3
-0.2
-0.6
-0.8
-1
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
3
w ( t ) y3 ( t ) A 3 sin 3 t 3 sin t sin 3t
A
4
w(t) a1 cost b1 sin t
a1=0
1 3A3 A3
b1 sin t sin 3t sin t dt
4 4
>>syms tet;syms A;
>>b1=‘((3*A^3/4)*sin(tet)-A^3/4*sin(3*tet))*sin(tet)’;
>>int(b1,-pi,pi) 1 3 A 3 3A3
b1
4 4
N(A, ) NA
4M 4
A A
As 2 3As 2A 4K 0
1 NA G (s) 0 9 2 A 2 4A * 2A 4K
3A
s1, 2
4 K 2A 2A
1 0 9 2 A 2 8 2 A 2 16KA
A s 3 s 2
2
s1, 2 1.5
2A
2 A 2 16KA
s1, 2 1.5
2A
Since there is always a negative real part, the system doesn’t exhibit a limit
cycle.
LYAPUNOV STABILITY FOR LINEAR TIME-INVARIANT SYSTEMS:
x A x
Let us consider a quadratic Lyapunov function candidate
V x T Px
where P is a given symmetric positive definite matrix.
Differentiating the positive definite function V along the system trajectory yields
another quadratic form
x T Px x T Px
V
where
V Ax Px x T PAx
T
x T A T Px x T P A x
x T ATP P A x A T P P A Q
x TQ x
If there exists a positive definite matrix Q satisfying the equation (Lyapunov
equation), the system is said to be stable in the sense of Lyapunov (ISL).
AT P P A Q 0 Lyapunov equation.
A useful way of studying a given linear system using scalar quadratic functions
is to derive a positive definite matrix P from a given positive definite matrix Q,
i.e.,
If P is positive definite, then xTPx is a Lyapunov function for the linear system
and global asymptotical stability is guaranteed.
Example:
Consider two matrices,
0 1 1 0
A ,Q
12 8 0 1
The linear system is stable (Real parts of all eigenvalues of the system matrix A
are negative) if there is a positive definite matrix P.
P=
clc;clear; 0.4010 -0.5000
A=[0 1;-12 -8]; -0.5000 0.8125
Q=[1 0;0 1];
P=lyap(A,Q) ans =
eig(P) 0.0661
1.1474
The matrix P is positive definite, since the eigenvalues are
real, and the system is stable ISL.
LYAPUNOV FUNCTION FOR NONLINEAR SYSTEM:
f
J(x )
x
If the matrix F=J+JT is negative definite, the equilibrium point at the origin is
asymptotically stable. A Lyapunov function for this system is
x 1 6 x1 2 x 2
x 2 2x1 6x 2 2x 32
We have
f1 f1
f x1 x 2 6 2
J
x f 2 f 2 2 6 6x 22
x1 x 2
12 4
FJJ T
2
4 12 12 x 2
The matrix F is negative definite over the whole state space. Therefore, the
origin is asymptotically stable, and a Lyapunov function candidate is
-8
clc;clear; -8.5
x2=-10:0.1:10; -9
2
-10
eg=eig(F)
-10.5
plot(eg(1),eg(2))
hold on -11
end -11.5
-12
-1400 -1200 -1000 -800 -600 -400 -200 0
1
6x1 2x 2
T
V( x ) f ( x )f ( x) 6x1 2x 2 , 2x1 6x 2 2x 3
2
2x 6x 2x 3
1 2 2
x 1 2x1
x 2 2x 2 2x1x 22
We assume that the gradient of the undetermined Lyapunov function has the
following form
a11=a22=1, a12=a21=0
which leads to
V1 x1 V2 x 2
Then
V x 2x 2 2x 2 1 x x
V 1 2 1 2
Thus, dV/dt is locally negative definite in the region (1-x1x2)>0. the function V
can be computed as
x12 x 22
x1 x2
V( x ) x1dx1 x 2dx 2
0 0
2
This is indeed positive definite, and therefore the asymptotic stability is
guaranteed.
Slotine and Li, Applied Nonlinear Control