Defining e x x0
The perturbation system
e x f ( x ) f (e x0 ) f1 (e)
e = 0 is an equilibrium point of the perturbation system.
We will always assume that 0 is the equilibrium point of study.
Norms
The Euclidean norm of an n-dimensional vector x, is defined as
1/ 2
n
2
L2 norm: x 2 xi
i 1
A norm is a measure of the size of x.
A norm must satisfy the following properties:
L∞ norm:
x
max | xi |
i 1, , n
x 2 1
x 1 1 x
1
x
x2 n x
x
x1n x
x2 2
x 2
x 2 1
1
1 1
x 1
1
x 1 x
1
Implications: If x
0 then x 1 0 and x 2 0
Notation: Ballr x : x r
x x
x x
x x3 x x3
x x
x x
x sin x x x4
Lyapunov stability means that the maximum deviation of the whole
trajectory from 0 depends continuously on the deviation of the
initial state.
R 0, r 0 such that x(0) Ballr x(t ) BallR , t
x(t)
Answer: R
Start in Challenge: reduced R
smaller r
t
r
Examples
For linear systems, all stable and marginally stable systems are
stable in the Lyapunov sense.
A Mechanical System: No friction x x 0
State Variables: x1 x, x2 x
State Space Model:
x1 x2 , x2 x1 x2
x(t)
Example: x
t
x x 2 x
Lyapunov stability is different from the condition
“bounded initial condition implies bounded state trajectories”
mx 2c( x 1) x kx 0
2
Van del Pol Equation
Limit Cycle:
x sin x
A system with multiple equilibrium points cannot be globally
asymptotically stable.
Local instability does not imply that trajectories will go to infinity.
They may go to other stable equilibrium points or limit cycles.
Lyapunov Linearization Methods
x f ( x) (1)
with f(0) = 0.
Assume that the nonlinear function f(x) is continuously differentiable
(so that partial derivatives which we are going to use are well
defined).
Jacobian Linearization
x Ax (2)
where A is the Jacobian matrix of f(x) at 0
f ( x )
A
x x 0
Lyapunov Stability: Local Linearization
Theorem (Lyapunov Linearization Method for Isolated
Equilibrium Points)
If the linear system (2) has poles on the imaginary axis and has no
poles in the right half plane, then no conclusion can be reached
about the equilibrium point.
Advantages of this theorem:
1. It is convenient to apply.
2. It is general, regardless the structure of the nonlinear
function f(x).
3. It implies a local design method: Locally linearize the
plant; design a linear controller that stabilizes the
linearized plant; this will lead to a locally stable
closed-loop system.
x bx 5
Unstable Node
Lyapunov Direct Methods
Lyapunov Direct Methods
Main Ideas:
1. It is usually difficult to solve a nonlinear differential
equation.
mx bx | x | k0 x k1 x3 0
x1 x, x2 x
x1 x2
b k0 k1 3
x2 x2 | x2 | x1 x1
m m m
Linearized System at 0:
x1 x2 0 1
Poles on the imaginary axis:
k0 A k0
x2 x1 0 Inconclusive!
m m
Total Mechanical Energy of the System:
V ( x, x) mx k0 x k1 x 3 dx
1 2 x
2 0
1 2 1 1 4
mx k0 x k1 x
2
2 2 4
Some properties of the energy:
V ( x, x) is continuously differentiable
V (0,0) 0 no energy at the equilibrium point
V ( x, x) 0 if ( x, x) 0 non-zero energy nearby
dV ( x1 , x2 ) V V x1
x b x2 0
3
,
dt x1 x2 2
total energy is not increasing along the system trajectories
Positive and Negative Definite Functions
Definition: Positive Definite Functions
A scalar continuous function V(x) of n-dimensional vector x is
said to be locally positive definite if
V(0)= 0
There exists R 0 such that
0
x R0 , x 0 V ( x ) 0
V ( x, x) is continuously differentiable
V (0,0) 0
V ( x, x) 0 if ( x, x) 0
dV ( x1 , x2 ) V V x1
2 0
3
, x b x
dt x1 x2 2
Theorem: Lyapunov Stability
Define 1 2
V ( x) (1 cos x1 ) x2
2
It is locally positive definite
dV ( x)
x2 sin x1 x2 x2 x22 0
dt
So, the equilibrium point is Lyapunov stable.
Main Ideas of the Proof
V(x)
For trajectories to
go beyond [-R,R], V(x)
must increase to pass this line
Smallest v(x)
on the boundary
dV ( x) x
0
dt will [-r,r]
make sure that
answer
this will not happen
[-R,R]
challenge
Illustration on the Phase Plane
x2
level sets of v(x)
V(x)=v1 ball of radius R
V(x)=v2 < v1
x1
dV ( x )
(b) dt
0 along any trajectories x of the system
in x R0 and x 0
Theorem: Asymptotic Lyapunov Stability
If for a constant R0 0, there exists a strict Lyapunov
function V(x) for the system at 0, then the
equilibrium point x = 0 is asymptotically Lyapunov
stable.
Basic Ideas of the Proof on Attraction (State Convergence)
x2
level sets of V(x)
V(x)=v1
V(x)=v2 < v1
x1
V(x)=v3 < v2
local region
of attraction:
ball of radius R0
dV ( x)
0 and V ( x) 0
dt x(t ) 0, t
implies V ( x(t )) C 0
C 0, since V ( x) is negative definite
Examples
x g ( x),
g (0) 0, xg ( x) 0 for x 0 and | x | r
Equilibrium point: x = 0
x
Define: V ( x) 0
g ( s) ds
0
V (0) g ( s) ds 0
0 Lyapunov Candidate
V ( x) 0, for x 0 and | x | r
dV ( x)
g ( x) x ( g ( x)) 2 0, for x 0 and | x | r
dt
This is a strict Lyapunov function at x = 0.
So, x = 0 is asymptotically stable.
Examples of such g(x)
x x
x x n , n is an odd integer
x
x sin x
x x cos x x
x x3e 2 x
x g ( x)
xg ( x) 0 simply implies:
x 0 g ( x) 0 x g ( x) 0 x decreases
x 0 g ( x) 0 x g ( x) 0 x increases
Example
Lyapunov candidate
(c) V ( x) as x
Theorem: Global Asymptotic Lyapunov Stability
V(x)
V(x) decreases
dV ( x)
2 x( x5 sin 6 x) 0, x 0
dt
V ( x) x 2 , x
Yes!!
Fact 1: If for a given positive definite matrix Q > 0 the
following matrix linear equation, called Lyapunov Equation, has
a positive definite solution P > 0, then x = 0 of the linear system
is globally asymptotically stable in the Lyapunov sense.
A P PA Q
T
Proof:
Define V ( x) xT Px 0, x 0, this is a Lyapunov candidate
V xT Px xT Px xT AT Px xT PAx
xT Qx 0, x 0
V ( x) , x
AT P PA Q
Robustness Analysis
x Ax ( x), ( x) is a small nonlinear perturbation and (0) 0
Suppose that A is stable. Then, there is P > 0 such that
AT P PA I
Define V ( x) xT Px 0
V xT Px xT Px
( xT AT T ( x)) Px xT P( Ax ( x))
xT x 2 xT P ( x)
If the second term is sufficiently small:
( x) x 2 xT P ( x) 2 k x , where k max ( P)
2
dV ( x)
2 k 1 0, x 0
dt
x = 0 is asymptotically stable,
robustly for any small perturbations.
Recall: Local Linearization Method
x f ( x), f (0) 0
Local linearization:
x Ax ( x),
A the Jacobian matrix
( x) higher order terms
( x)
f ( x) is continuously differentable 0,x 0
x
x Ax ( x),
( x) is a small nonlinear perturbation and (0) 0
Construction of Lyapunov Functions
x1 x2
b k0 k
x2 x2 | x2 | x1 1 x13
m m m
1 2 1 1 4
V ( x1 , x2 ) mx2 k0 x1 k1 x1
2
2 2 4
Try Some Functions of Norms of the State:
x x5 sin 6 x, V ( x) x 2
Lyapunov candidate
Krasovski's Methods
3 1 6 2
1 1 3x 2 2 2 6x 2
2 2
Invariant Set Theorem
Motivations
Issue:
Lyapunov functions we construct often can only conclude
Lyapunov stability, rather than asymptotic stability. Typical
examples are energy functions. Searching for strict
Lyapunov functions can be very difficult.
Question:
Can we use the Lyapunov functions to study asymptotic
stability also?
Example:Energy of a Mechanical System
mx bx | x | k0 x k1 x3 0
x1 x, x2 x
x1 x2
b k0 k1 3
x2 x2 | x2 | x1 x1
m m m
Linearized System at 0:
x1 x2 0 1
Poles on the imaginary axis:
k0 A k0
x2 x1 0 Inconclusive!
m m
Total Mechanical Energy of the System:
V ( x, x) mx k0 x k1 x 3 dx
1 2 x
2 0
1 2 1 1 4
mx k0 x k1 x
2
2 2 4
Some properties of the energy:
V ( x, x) is continuous
V (0,0) 0 no energy at the equilibrium point
V ( x, x) 0 if ( x, x) 0 non-zero energy nearby
dV ( x1 , x2 ) V V x1
x b x2 0
3
,
dt x1 x2 2
total energy is not increasing along the system trajectories
Issue:
Lyapunov functions work for isolated equilibrium points.
Question:
How can we study stability of limit cycles and continuous
equilibrium points?
Issue:
For practical purposes, it is important to obtain regions of
contraction (how close is close enough for the system to
maintain local stability?). Lyapunov functions may not be the
best functions to obtain such regions.
Invariant Sets
x(t0 ) G x(t ) G, t t0
Typical Invariant Sets
Equilibrium points.
x2
0 0
x x
0 1 x1
Limit cycles.
If V(x) is a Lyapunov function of the system x f ( x ) , then
the level set
l {x : V ( x) l} note: V ( x) 0 in l
is an invariant set of the system.
x1 x2
b k k
x2 x2 | x2 | 0 x1 1 x13
m m m
V ( x, x) mx k0 x k1 x 3 dx
1 2 x
2 0
1 1 1
mx 2 k0 x 2 k1 x 4
2 2 4
100 ( x, x) : V ( x, x) 100
(the set of the states that carry energy no more than 100)
Invariant Set Theorems
dV ( x1 , x2 ) V V x1
x b x2 0 x2 0
3
,
dt x1 x2 2
100 ( x, x) : V ( x, x) 100
x1
R
At a point in R other than (0,0):
M x1 0
k0 k1 3 it will leave R
x2 x1 x1
m m
xb( x ) 0, for x 0
xc( x ) 0, for x 0
x1 x2
x1 x, x2 x
x2 b( x2 ) c( x1 )
1 2 x
V ( x, x) x c( y)dy
2 0
V ( x, x) is continuous
V (0,0) 0
V ( x, x) 0 if ( x, x) 0
R ( x,0) : V ( x,0) L
M (0, 0)
1 2
x2
x14 2 x22 10 is a limit cycle
V ( x 2 x 10)
4
1
2
2
2
dV
8( x14 3x26 )( x14 2 x22 10) 2 0
dt
dV
8( x14 3x26 )( x14 2 x22 10) 2 0
dt
(0, 0) (the equilibrium point)
x14 2 x22 10 (the limit cycle)
20 ( x1 , x2 ) : V ( x1 , x2 ) ( x14 2 x22 10)2 20
20
A Simple Adaptive Control System
Plant: y ay u, a is unknown
Controller: u ky
Adaptation Law: k y 2
k
Overall System
y ay ky
k y 2
y
1 2 1
V y ( k b) 2
2 2
dV 1
yy (k b)k (b a) y 2 0, if b a
dt
V 0
1 2 1
L ( y , k ) : y ( k b) L
2
2 2
R ( y, k ) : V 0 L (0, k ) : (k b) L
1 2
2
M R
k L
By the Invariant Set Theorem:
Starting at any x(0) L ,
x(t ) M , t
M R
y0
y
The output will go to zero;
The gain k will approach some constant