Chapter 5.

Introduction to Nonlinear Systems

Modern Control Theory (Course Code: 10213403)

Professor Jun WANG (

Department of Control Science & Engineering School of Electronic & Information Engineering Tongji University

Spring semester, 2012

Outline
1

5.1 Introduction

5.2 Common nonlinear elements

5.3 Properties of nonlinear systems

5.4 Approaches to analyzing nonlinear systems

5.5 Summary

Outline
1

5.1 Introduction

5.2 Common nonlinear elements

5.3 Properties of nonlinear systems

5.4 Approaches to analyzing nonlinear systems

5.5 Summary

5.1 Introduction

Linear and nonlinear systems

Main features of linear systems

Principle of superposition c1 (t) = f r1 (t) , c2 (t) = f r2 (t) c1 (t) + c2 (t) = f r1 (t) + r2 (t)

Typical inputs are used to generate transfer functions. Complete set of mathematical tools is available for analysis and design: e.g. ODE, Laplace transformation, etc.

Main features of nonlinear systems

Principle of superposition is not applicable No unied solution

Prof J Wang (Tongji Uni)

Chap 5. Nonlinear system introduction

Spring 2012

4 / 24

5.1 Introduction

Why to study nonlinearity?

Most practical systems are nonlinear

All linear systems we study are only the approximations to the practical systems to certain extent. For some systems, the nonlinearities are not negligible.

Special result may be achieved by using nonlinear control. e.g. Posicast control u(t)/y(t) u(t)/y(t)

1.5 1.0 0.5

1.5 1.0 0.5

Otto Smith
(1917-2008)
Prof J Wang (Tongji Uni)

10

15

20

10

15

20

t
5 / 24

Chap 5. Nonlinear system introduction

Spring 2012

Outline
1

5.1 Introduction

5.2 Common nonlinear elements

5.3 Properties of nonlinear systems

5.4 Approaches to analyzing nonlinear systems

5.5 Summary

5.2 Common nonlinear elements

Some typical nonlinearities

y y y

k x a a x k a

k a x

A nonlinear spring y = k1 x + k2 x3 , k2 > 0, k2 > 0

Prof J Wang (Tongji Uni)

Saturation e.g. Electronic amp. e.g. Power limitation in servo motors

Chap 5. Nonlinear system introduction

Dead zone e.g. Relay amp. e.g. Actuator

Spring 2012

7 / 24

5.2 Common nonlinear elements

Some typical nonlinearities (cont.)

y M y y

M Idea reley Hysteresis Backlash

Prof J Wang (Tongji Uni)

Chap 5. Nonlinear system introduction

Spring 2012

8 / 24

Outline
1

5.1 Introduction

5.2 Common nonlinear elements

5.3 Properties of nonlinear systems

5.4 Approaches to analyzing nonlinear systems

5.5 Summary

5.3 Properties of nonlinear systems

Features of nonlinear systems

Nonlinearity has some distinguished properties dierent from the linearity. Multiple equilibrium points Self-excited oscillations (limit cycles) Bifurcations Chaos ...

Prof J Wang (Tongji Uni)

Chap 5. Nonlinear system introduction

Spring 2012

10 / 24

5.3 Properties of nonlinear systems

Multiple equilibrium points

Nonlinear systems often have more than one equilibrium points. The possible equilibrium point depends on the systems parameters, initial conditions and external excitations. Let us study this feature by an example Example Consider the rst order system (t) = x(t) + x2 (t) x compare its response with that of the linearized system subject to varied initial conditions x0 .
Prof J Wang (Tongji Uni) Chap 5. Nonlinear system introduction Spring 2012 11 / 24

5.3 Properties of nonlinear systems

Solutions The linearized model of the given system around the state x(t) = 0 is (t) = x(t) x and its response to any initial condition x0 is x(t) = x0 et x(t) What can be observe? The linearized system has a o t unique equilibrium point. The equilibrium point is x = 0

Prof J Wang (Tongji Uni)

Chap 5. Nonlinear system introduction

Spring 2012

12 / 24

5.3 Properties of nonlinear systems

The response of the nonlinear system is x(t) = x0 et 1 x0 + x0 et

Depending on initial conditions, the trajectory will end in one of the two possible equilibrium points x = 0 and x = 1. x(t) For x0 > 1 limt x(t) = For x0 = 1 x(t) 1 1 o t For x0 < 1 limt x(t) = 0 Hence, the system has two equilibrium points xe = 0 and xe = 1, which depend on the initial condition x0 .
Prof J Wang (Tongji Uni) Chap 5. Nonlinear system introduction Spring 2012 13 / 24

5.3 Properties of nonlinear systems

Self-excited oscillations (limit cycles)

What are self-excited oscillations or limit cycles? Nonlinear systems possibly have oscillations with amplitude and frequency independent of the initial conditions of the system. (Question: How about linear system?) The occurrence of the oscillations relies on the initial conditions. This phenomenon can be illustrated by the Van der Pol equation. Example (Van der Pol Equation) Consider the second-order nonlinear dierential equation (t) + 2c x2 (t) 1 x (t) + kx(t) mx where m, c and k are positive constants.
Prof J Wang (Tongji Uni) Chap 5. Nonlinear system introduction Spring 2012 14 / 24

5.3 Properties of nonlinear systems

Solutions (t) + 2c x2 (t) 1 x (t) + kx(t) look like? What does the equation mx |x(t)| > 1 system stable system energy |x(t)| |x(t)| < 1 system unstable system energy |x(t)| Therefore, the system motion can neither grow unboundedly or decay to zero. x(t)
3 2 1 2 1 2 3 4 6 8 10

The responses of the system with dierent initial conditions display sustained oscillation with t the same amplitude and frequency.

Prof J Wang (Tongji Uni)

Chap 5. Nonlinear system introduction

Spring 2012

15 / 24

5.3 Properties of nonlinear systems

Chaos

)? What is chaos ( In Greek mythical cosmogony, particularly in the Theogony (Origin of the Gods) of Hesiod (

), Chaos is the

original dark void from which everything else appeared.

Hesiod and the Muse by Gustave Moreau

Prof J Wang (Tongji Uni)

Chap 5. Nonlinear system introduction

Spring 2012

16 / 24

Vortex street in the atmosphere

Edward Lorenz The discoverer of Chaos

Chaos James Gleick

5.3 Properties of nonlinear systems

Concept of Chaos
Some observations For stable linear systems, small variations in initial conditions will lead to small variations in the output. For nonlinear systems, small variations in initial conditions can lead to huge variations in the output. Denition A commonly used denition says that, for a dynamical system to be classied as chaotic, it must have the following properties:
1 2 3

It must be sensitive to initial conditions; It must be topologically mixing; and Its periodic orbits must be dense.
Prof J Wang (Tongji Uni) Chap 5. Nonlinear system introduction Spring 2012 18 / 24

Example Let us consider the following simple nonlinear system + 0.1x + x5 = 6 sin t x (0) = 3 and with the two sets of initial conditions x(0) = 2, x (0) = 3.01. x(0) = 2.01, x Solutions The response of the system subject to initial conditions is shown below x(t)
2 1 5 1 2 10 15 20 25 30 35 40 45 50

Outline
1

5.1 Introduction

5.2 Common nonlinear elements

5.3 Properties of nonlinear systems

5.4 Approaches to analyzing nonlinear systems

5.5 Summary

5.4 Approaches to analyzing nonlinear systems

How to study nonlinear control systems?

Can we linearize some nonlinear systems?

YES! Analysis by linearization

Can we extend the analysis and synthesis methods of linear systems to some nonlinear ones?

YES! Root locus equivalent gain analysis using the root locus YES! Frequency response equivalent gain analysis using frequency response

Any methods dealing with nonlinear behavior directly?

YES! Phase plane method (graphical representation of ODEs) YES! Lyapunov methods YES! ... ...
Chap 5. Nonlinear system introduction Spring 2012 21 / 24

Prof J Wang (Tongji Uni)

Outline
1

5.1 Introduction

5.2 Common nonlinear elements

5.3 Properties of nonlinear systems

5.4 Approaches to analyzing nonlinear systems

5.5 Summary

5.5 Summary

How to study nonlinear control systems?

Almost everything is nonlinear! There is NO unied methods for nonlinear system analysis and synthesis! Nonlinear control is still an active and developing researching eld! In this course, we will study three fundamental methods for nonlinear systems

Describing functions (equivalent frequency response) Phase plane method (graphical representation of state space) Lyapunov methods

Any further interest? Please read from the famous book Applied Nonlinear Control by Slotine and Li.
Prof J Wang (Tongji Uni) Chap 5. Nonlinear system introduction Spring 2012 23 / 24

Chapter 5. Introduction to Nonlinear Systems

Modern Control Theory (Course Code: 10213403)

Professor Jun WANG (

Department of Control Science & Engineering School of Electronic & Information Engineering Tongji University

Spring semester, 2012

Go to next chapter!