Nonlinear Control

2012

Office Room:
Telephone: Ext. 3221
E-mail: shinylin@mail.cgu.edu.tw
Office Hour: 2:00 4:00 pm, Friday


Textbook:
Jean-Jacques E. Slotine and Weiping Li, Applied
Nonlinear Control, Pearson Education Taiwan
Ltd., 1991.

Reference:
Alberto Isidori, Nonlinear Control Systems,
Springer-Verlag, 1999.


1. Phase Portrait Lyapunov
Method
2. Feedback Linearization, Sliding
Control Adaptive Control

Linear System Theory
Elementary Differential Equations


(20%)
2 ( 40%)

Chapter 1

Introduction

1.1 Why Nonlinear Control

A: To control nonlinear systems.

 Improvement of Existing Control Systems

- Linear control methods rely on the key
assumption of small range operation for the
linear model to be valid.
- Nonlinear controllers may handle the
nonlinearities in large range operation directly,
because the controller is designed for handling
the nonlinear system directly.

 Analysis of hard nonlinearities

-Linear control assumes the system model
is linearizable.
-Hard nonlinearities: nonlinearities whose
discontinuous nature does not allow linear
approximation.
-Coulomb friction, saturation, dead-zones,
backlash, and hysteresis.

 Dealing with Model Uncertainties

- In designing linear controllers, we assume that
the parameters of the system model are
reasonably well known.
- In real world, control problems involve
uncertainties in the model parameters.
- The model uncertainties can be tolerated in
nonlinear control, because the uncertainty is
taken into account in the controller design.

 Design Simplicity
-Good nonlinear controller designs may be
simpler and more intuitive than their linear
counterparts.
-This result comes from the fact that
nonlinear controller designs are often
deeply rooted in the physics of the plants.
-Example: pendulum

1.2 Nonlinear System Behavior

Nonlinearities
- Inherent (natural) : Coulomb friction
between contacting surfaces.
- Intentional (artificial): adaptive control laws.
- Continuous
- Discontinuous: Hard nonlinearities
(backlash, hysteresis) cannot be locally
approximated by linear function.

Linear Systems
Linear time-invariant (LTI) control systems, of the
form

x& Ax

with x being a vector of states and A being the

system matrix.

Properties of LTI systems

Unique equilibrium point if A is nonsingular
Stable if all eigenvalues of A have negative
real parts, regardless of initial conditions
General solution can be solved analytically

Common Nonlinear System Behaviors

I. Multiple Equilibrium Points
Nonlinear systems frequently have more than
one equilibrium point (an equilibrium point is
a point where the system can stay forever
without moving, i.e. a point where x& 0 ).

Example 1.2: A first-order system

x& x x

Its linearization around

x(t ) 0is

x& x
with solution x(t) = x(0)e t : general solution
can be solved analytically.
-Unique equilibrium point at x = 0.
-Stable regardless of initial condition.

- Integrating equation dx/( x + x2)=dt

t

x0 e
x(t )
t
1 x0 x0 e
-Tow equilibrium points, x = 0 and x = 1.
- Qualitative behavior strongly depends on its
initial condition.

Figure 3.1: Responses of the linearized

system (a) and the nonlinear system (b)

Stability of Nonlinear Systems May

Depend on Initial Conditions:
- Motions starting with x0 1 converges.
- Motions starting with x0 > 1 diverges.

Properties of LTI Systems:

In the presence of an external input u(t), i. e.,
with
x& Ax Bu
-Principle of superposition.
-Asymptotic stability implied BIBO stability
in the presence of u.

Stability of Nonlinear Systems May

Depend on Input Values:
A bilinear system

x& xu

u 1 , converges.
u 1 , diverges.

II. Limit Cycles

-Oscillations of fixed amplitude and fixed
period without external excitation.

Example 1.3: Van der Pol Equation

& 2c( x 1) x& kx 0

mx&
2

where m, c and k are positive constants.

- A mass-spring-damper system with a

position-dependent damping coefficient
2c (x2-1)
- For large x, 2c (x2-1)>0 : the damper removes
energy from the system - convergent
tendency.
- For small x, 2c (x2-1)<0 : the damper adds
energy to the system - divergent tendency.

-Neither grow unboundedly nor decay to zero.

- Oscillate independent of initial conditions.

Figure 2.8:Phase portrait of the Van der Pol equation

- Limit cycle (case for m=1, c=1 and k=1)

The trajectories starting from both outside and inside
converge to this curve.

II. Limit Cycles (continued)

-Oscillations of fixed amplitude and fixed
period without external excitation.
Example 1.4:

&
x& ( x x& 1) x& x 0
2

Common Nonlinear System Behaviors

III. Bifurcations
-As parameters changed, the stability of the
equilibrium point can change.
-critical or bifurcation values :
Values of the parameters at which the
qualitative nature of the systems motion
changes.

-Topic of bifurcation theory: Quantitative

change of parameters leading to qualitative
change of system properties.
- Undamped Duffing equation

&
x& Equation
x x is0
(the damped Duffing
3

3
, which
may
represent
a
mass-damper-spring
system
&
x& cx& x x 0
with a hardening spring).

- As varies from + to -, one equilibrium point splits into

3 points (
), as shown in Figure 1.5(a).
xe , , 0
is a critical bifurcation value.

Figure 1.5: (a) a pitchfork bifurcation

(b) a Hopf bifurcation

Common Nonlinear System Behaviors

IV. Chaos
-The system output is extremely sensitive to
initial conditions.
-Essential feature: the unpredictability of the
system output.

Simple Nonlinear system

&
x& 0.1x& x 6sin t
-Two almost identical initial conditions,
Namely x(0) 2, x&(0) 3 , and
x(0) 2.01, x&(0) 3.01.
- The two responses are radically different
after some time.
5

Figure 1.6: Chaotic behavior of a nonlinear

system

Outlines of this Course

I. Phase plane analysis
II. Lyapunov theory
III. Feedback linearization
IV. Sliding control
VI. Adaptive control