ET301

Control Systems
Compiled & Edited by,

Dr. Aniket D. Gundecha©

Assistant Professor

School of Electrical Engineering

Alandi (D), Pune.

 THE FUTURE BELONGS TO THOSE
WHO BELIEVE IN
THE BEAUTY OF THEIR DREAMS

Dr. Aniket D. Gundecha©2020 , MIT AOE, Pune, INDIA. All rights reserved

No part of this publication may be reproduced, stored in retrieval system, or transmitted, in

any form, or by any means, electronic, mechanical, photocopying, recording or otherwise,
without prior permission in writing from the author. Although every effort has been made
to ensure that the material provided is error free, the author shall not be held responsible
or liable for any damages.
Contents

List of Figures xii

List of Tables xv

1 Modeling in Frequency Domain 1

1.1 Introduction to Control Systems . . . . . . . . . . . . . . . . . . . . 2

1.1.1 What is a System? . . . . . . . . . . . . . . . . . . . . . . . 2

1.1.2 Examples of System . . . . . . . . . . . . . . . . . . . . . . 2

1.2 Classification of Systems . . . . . . . . . . . . . . . . . . . . . . . . 3

1.2.1 Linear Vs Non-Linear Systems . . . . . . . . . . . . . . . . . 3

1.2.2 Static Vs Dynamic Systems . . . . . . . . . . . . . . . . . . 4

1.2.3 Time Invarient Vs Time Varient Systems . . . . . . . . . . . 4

1.2.4 Causal Vs Non-causal Systems . . . . . . . . . . . . . . . . . 4

1.3 What is a Control System? . . . . . . . . . . . . . . . . . . . . . . . 5

1.3.1 Control System . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.3.2 Control System with disturbance . . . . . . . . . . . . . . . 5

1.3.3 Feedback in Control . . . . . . . . . . . . . . . . . . . . . . 6

1.4 Examples of Control Systems . . . . . . . . . . . . . . . . . . . . . 6

1.4.1 Open-Loop Control system (Example–1) . . . . . . . . . . . 6

v
 Contents

1.4.2 Open-Loop Control system (Example–2) . . . . . . . . . . . 7

1.4.3 Closed-Loop (feedback) Control system (Example–1) . . . . 7

1.4.4 Closed-Loop (feedback) Control system (Example–2) . . . . 7

1.5 Terminologies used in Control Systems . . . . . . . . . . . . . . . . 8

1.5.1 Elements of Control Systems . . . . . . . . . . . . . . . . . . 8

1.6 The Roadmap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.6.1 The control system design process . . . . . . . . . . . . . . . 8

1.7 Modelling of Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.7.1 What is a Model? . . . . . . . . . . . . . . . . . . . . . . . . 9

1.8 Model Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.8.1 First Principles (Analytical modelling) . . . . . . . . . . . . 9

1.8.2 Empirical Data (Experimental modelling) . . . . . . . . . . 10

1.9 Types of Mathematical Models . . . . . . . . . . . . . . . . . . . . 11

1.9.1 Examples of Modelling . . . . . . . . . . . . . . . . . . . . . 11

1.10 Modelling Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.10.1 Steps of Analytical Modelling . . . . . . . . . . . . . . . . . 12

1.11 Elements of Modelling . . . . . . . . . . . . . . . . . . . . . . . . . 13

1.11.1 Physical Systems . . . . . . . . . . . . . . . . . . . . . . . . 13

1.12 Electrical Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.12.1 Types of sources . . . . . . . . . . . . . . . . . . . . . . . . 14

1.12.2 Types of sources . . . . . . . . . . . . . . . . . . . . . . . . 14

1.12.3 Elements of Electrical Systems . . . . . . . . . . . . . . . . . 15

1.12.4 Elements of Electrical Systems . . . . . . . . . . . . . . . . . 15

1.13 Analysis of Electrical Systems . . . . . . . . . . . . . . . . . . . . . 16

1.13.1 Nodal and Loop Analysis . . . . . . . . . . . . . . . . . . . . 16

vi
 Contents

1.13.2 Nodal Analysis–Example . . . . . . . . . . . . . . . . . . . . 17

1.13.3 Mesh Analysis–Example . . . . . . . . . . . . . . . . . . . . 17

1.14 Mechanical Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1.14.1 Classification . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1.14.2 Elements of Translational Systems . . . . . . . . . . . . . . 18

1.14.3 Units for Translational Systems . . . . . . . . . . . . . . . . 20

1.14.4 Example of Spring Mass Damper . . . . . . . . . . . . . . . 20

1.14.5 Freebody Diagram . . . . . . . . . . . . . . . . . . . . . . . 21

1.14.6 Free-body diagram–Example . . . . . . . . . . . . . . . . . . 21

1.15 Electrical Analogies of Mechanical Systems . . . . . . . . . . . . . . 22

1.15.1 Force Voltage Analogy . . . . . . . . . . . . . . . . . . . . . 22

1.15.2 Force Current Analogy . . . . . . . . . . . . . . . . . . . . . 22

1.16 Transfer Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

1.16.1 Need of Transfer Function . . . . . . . . . . . . . . . . . . . 23

1.16.2 What is a Transfer Function? . . . . . . . . . . . . . . . . . 23

1.16.3 Impulse response . . . . . . . . . . . . . . . . . . . . . . . . 23

1.17 Transfer Function Steps . . . . . . . . . . . . . . . . . . . . . . . . 24

1.17.1 Steps to find Transfer function . . . . . . . . . . . . . . . . . 24

1.17.2 Properties of Transfer Function . . . . . . . . . . . . . . . . 26

1.18 Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

1.18.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

1.18.2 Block Diagram of a System . . . . . . . . . . . . . . . . . . 26

1.19 Components of Block Diagram . . . . . . . . . . . . . . . . . . . . . 27

1.19.1 Basic Elements of Block Diagram . . . . . . . . . . . . . . . 27

1.19.2 Basic Elements of Block Diagram . . . . . . . . . . . . . . . 27

vii
 Contents

1.20 Developing Block Diagram from Mathematical model . . . . . . . . 28

1.20.1 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.21 Block Diagram Algebra . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.21.1 Basic connections for Blocks . . . . . . . . . . . . . . . . . . 28

1.22 Block Diagram Reduction . . . . . . . . . . . . . . . . . . . . . . . 30

1.22.1 Block Diagram Algebra for Summing Points . . . . . . . . . 30

1.22.2 Block Diagram Algebra for Take-off Points . . . . . . . . . . 31

1.22.3 Block Diagram Reduction Rules . . . . . . . . . . . . . . . . 31

1.23 Signal Flow Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

1.23.1 What is Signal Flow Graph? . . . . . . . . . . . . . . . . . . 33

1.23.2 The Difference . . . . . . . . . . . . . . . . . . . . . . . . . . 33

1.23.3 Properties of Signal Flow Graphs . . . . . . . . . . . . . . . 33

1.24 Basic Elements of Signal Flow Graph . . . . . . . . . . . . . . . . . 34

1.24.1 Basic Elements . . . . . . . . . . . . . . . . . . . . . . . . . 34

1.25 Procedure to draw signal flow graph . . . . . . . . . . . . . . . . . . 36

1.25.1 Signal flow graph from Block diagram . . . . . . . . . . . . . 36

1.25.2 Mason’s Gain formula . . . . . . . . . . . . . . . . . . . . . 36

2 Modeling in Time Domain 41

2.1 Introduction to State Space systems . . . . . . . . . . . . . . . . . . 42

2.1.1 Transfer function modelling approach . . . . . . . . . . . . . 42

2.1.2 Advantages of state space method . . . . . . . . . . . . . . . 42

2.2 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

2.2.1 Definitions related to state space techniques . . . . . . . . . 43

2.3 Mathematical Representation of state space . . . . . . . . . . . . . 44

viii
 Contents

2.3.1 State equation . . . . . . . . . . . . . . . . . . . . . . . . . . 44

2.3.2 Output equation . . . . . . . . . . . . . . . . . . . . . . . . 44

2.4 State space equations – Matrix form . . . . . . . . . . . . . . . . . 44

2.4.1 State equation – Matrix form . . . . . . . . . . . . . . . . . 44

2.4.2 Output equation – Matrix form . . . . . . . . . . . . . . . . 45

2.4.3 State space Matrices . . . . . . . . . . . . . . . . . . . . . . 45

2.4.4 Block diagram representation of state space model . . . . . . 47

2.4.5 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

2.5 Selection of state variables . . . . . . . . . . . . . . . . . . . . . . . 48

2.6 Companion form (controllable canonical form) . . . . . . . . . . . . 49

2.6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

2.6.2 State space representation in controllable canonical form –

Case 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

2.6.3 Controllable canonical form – Case 2 . . . . . . . . . . . . . 51

2.7 Introduction to observable canonical form . . . . . . . . . . . . . . 54

2.7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

2.7.2 Mathematical representation of observable canonical form . . 55

2.8 Introduction to diagonal canonical form . . . . . . . . . . . . . . . . 57

2.9 Mathematical representation of diagonal canonical form . . . . . . . 58

2.9.1 Diagonal canonical form – Case 1 . . . . . . . . . . . . . . . 58

2.9.2 Block diagram representation of diagonal canonical form–Case

1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

2.9.3 Diagonal canonical form – Case 2 . . . . . . . . . . . . . . . 61

2.9.4 Block diagram representation of diagonal canonical form–Case

2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

2.9.5 Advantages of diagonal canonical form . . . . . . . . . . . . 65

ix
 Contents

2.9.6 Disadvantages of diagonal canonical form . . . . . . . . . . . 65

2.10 State space model to Transfer Function . . . . . . . . . . . . . . . . 66

2.10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

2.10.2 Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

2.10.3 Determination of Adj [sI-A] . . . . . . . . . . . . . . . . . . 67

2.10.4 How to find inverse of third order system? . . . . . . . . . . 67

2.10.5 How to find determinant of [A]? . . . . . . . . . . . . . . . . 68

2.11 Time-Domain Solution of LTI State Equations . . . . . . . . . . . . 70

2.11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

2.12 Solution of Homogeneous State Equations . . . . . . . . . . . . . . 70

2.12.1 Solution of the scalar differential equation . . . . . . . . . . 70

2.12.2 Solution of the vector-matrix differential equation . . . . . . 71

2.12.3 Laplace transform approach to solution–Scalar case . . . . . 73

2.12.4 Laplace transform approach to solution–Vector-matrix case . 73

2.12.5 State–Transition Matrix . . . . . . . . . . . . . . . . . . . . 74

2.12.6 Important points about State–Transition Matrix . . . . . . . 74

2.12.7 Properties of State–Transition Matrix . . . . . . . . . . . . . 75

2.13 Solution of Non-homogeneous State Equations . . . . . . . . . . . . 75

2.13.1 Solution of the scalar differential equation . . . . . . . . . . 75

2.13.2 Solution of the vector-matrix differential equation . . . . . . 76

2.13.3 Laplace transform approach to solution–Vector-matrix case . 77

2.14 Introduction to controllability and observability . . . . . . . . . . . 79

2.15 Controllability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

2.15.1 Controllability Matrix . . . . . . . . . . . . . . . . . . . . . 79

2.16 Observability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

x
 Contents

2.16.1 Observability Matrix . . . . . . . . . . . . . . . . . . . . . . 81

2.17 Introduction to Eigenvalues and Eigenvectors . . . . . . . . . . . . 82

2.17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

2.18 Eigenvectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

2.19 Eigenvalues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

2.19.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

2.19.2 Properties of Eigenvalues . . . . . . . . . . . . . . . . . . . . 83

xi
List of Figures

1.1 System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 System with subsystems . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Control System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.4 Control System with disturbance . . . . . . . . . . . . . . . . . . . 5

1.5 Feedback Control System with disturbance . . . . . . . . . . . . . . 6

1.6 Open-Loop Control system . . . . . . . . . . . . . . . . . . . . . . . 6

1.7 Open-Loop Control system . . . . . . . . . . . . . . . . . . . . . . . 7

1.8 Closed-Loop Control system . . . . . . . . . . . . . . . . . . . . . . 7

1.9 Closed-Loop Control system . . . . . . . . . . . . . . . . . . . . . . 7

1.10 Closed-Loop Control system . . . . . . . . . . . . . . . . . . . . . . 8

1.11 The control system design process . . . . . . . . . . . . . . . . . . . 8

1.12 From first principles (Analytical modelling)–Example . . . . . . . . 10

1.13 Empirical Data (Experimental modelling–Example . . . . . . . . . 10

1.14 Modelling steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.15 Automobile Suspension system . . . . . . . . . . . . . . . . . . . . 20

1.16 Original System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

1.17 Free-body diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

xii
 List of Figures

1.18 Electrical circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

1.19 Electrical circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

1.20 Mechanical system . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

1.21 Mechanical system . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

1.22 Block diagram of a closed loop system . . . . . . . . . . . . . . . . 27

1.23 Cascaded subsystems . . . . . . . . . . . . . . . . . . . . . . . . . . 29

1.24 Equivalent transfer function . . . . . . . . . . . . . . . . . . . . . . 29

1.25 Parallel subsystems . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

1.26 Equivalent transfer function . . . . . . . . . . . . . . . . . . . . . . 29

1.27 Moving a block to the left past a summing junction . . . . . . . . . 30

1.28 Moving a block to the right past a summing junction . . . . . . . . 30

1.29 Moving a block to the left past a pickoff point . . . . . . . . . . . . 31

1.30 Moving a block to the right past a pickoff point . . . . . . . . . . . 31

1.31 Block diagram of a system . . . . . . . . . . . . . . . . . . . . . . . 32

1.32 Block diagram of a system . . . . . . . . . . . . . . . . . . . . . . . 32

1.33 Block diagram of a system . . . . . . . . . . . . . . . . . . . . . . . 32

1.34 Signal Flow Graph of a system . . . . . . . . . . . . . . . . . . . . . 34

1.35 Signal flow graph of a system . . . . . . . . . . . . . . . . . . . . . 37

1.36 Signal flow graph of a system . . . . . . . . . . . . . . . . . . . . . 37

1.37 Signal flow graph of a system . . . . . . . . . . . . . . . . . . . . . 37

1.38 Signal flow graph of a system . . . . . . . . . . . . . . . . . . . . . 38

1.39 Block diagram of a system . . . . . . . . . . . . . . . . . . . . . . . 38

1.40 Lag network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

xiii
 List of Figures

1.41 Lead network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

1.42 Mechanical system . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

1.43 Mechanical system . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

1.44 Mechanical system . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

2.1 System block diagram . . . . . . . . . . . . . . . . . . . . . . . . . 43

2.2 MIMO system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

2.3 Block diagram representation of state space . . . . . . . . . . . . . 47

2.4 Block diagram of equation (2.20) . . . . . . . . . . . . . . . . . . . 55

2.5 Block diagram representation of diagonal canonical form–Case 1 . . 61

2.6 Block diagram representation of diagonal canonical form–Case 2 . . 64

xiv
List of Tables

1.1 Types of Mathematical Models . . . . . . . . . . . . . . . . . . . . 11

1.2 Variables of mechanical translational system . . . . . . . . . . . . . 20

xv
Chapter 1

Modeling in Frequency Domain

Learning Outcomes
After completing this chapter, students will be able to:

ˆ Explain types of control system

ˆ Illustrate feedback control system and its effect

ˆ Define Transfer Function, Characteristic Equation and pole-zero of system

ˆ Determine the transfer function of electrical and mechanical network

ˆ Apply block reduction technique to complex mathematical system

ˆ Apply Signal Flow Graph technique to complex mathematical system

1
 Chapter 1. Modeling in Frequency Domain

1.1 Introduction to Control Systems

1.1.1 What is a System?

ˆ A collection of physical, biological or abstract components which together

perform an intended objective.

ˆ A system gives an output (also called response) for an input (also called
excitation)

Input/Excitation Output/Response
System

Figure 1.1: System

Sys2
Input/Excitation Output/Response
Sys1 + Sys3

Sys4

Figure 1.2: System with subsystems

1.1.2 Examples of System

Motor

ˆ Input–Electrical energy (voltage)

ˆ Output–Mechanical energy (Torque/Rotation)

Air Conditioner

ˆ Input–Electrical energy (voltage)

ˆ Output–Heat energy (Changes the ambient Temp)

2
 Chapter 1. Modeling in Frequency Domain

Human body infected with a virus

ˆ Input–Drug administration

ˆ Output–Drug distribution and effect on the body

Vehicle (car or bus)

ˆ Input–Acceleration or Deceleration

ˆ Output–Vehicle displacement

1.2 Classification of Systems

ˆ Variety of classifications are possible based on system features and applications

ˆ Some of the important classifications include:

– Linear and non-linear systems

– Static and dynamic systems

– Time invarient and time varient systems

– Causal and non-causal systems

1.2.1 Linear Vs Non-Linear Systems

Linear Systems Non-Linear Systems

ˆ Output of the system varies linearly ˆ Output of the system does not vary
with input linearly with input

ˆ Satisfies homogeneity and ˆ Do not satisfy homogeneity and

superposition superposition
V
ˆ E.g. Resistor : I = V
R
ˆ E.g. Diode : I = I0 (e τ − 1)

3
 Chapter 1. Modeling in Frequency Domain

1.2.2 Static Vs Dynamic Systems

Static Systems Dynamic Systems

ˆ At any time, output of the system ˆ Output of the system depends on

depends only on present input present as well as past inputs

ˆ Memory less systems ˆ Presence of memory

ˆ y(t) = f (u(t)) ˆ y(t) = f (u(t), u(t − 1), ...)

Rt
ˆ E.g. Resistor : I = V
R
ˆ E.g. Inductor:I(t) = 1
L 0
V (t)dt

1.2.3 Time Invarient Vs Time Varient Systems

Time Invarient Systems Time Varient Systems

ˆ Output of the system is independent of ˆ Output of the system varies dependent

the time at which input is applied on the time at which input is applied

ˆ y(t) = f (u(t)) ⇒ y(t + δ) = f (u(t + δ)) ˆ y(t) = f (u(t)) ; y(t + δ) = f (u(t + δ))

ˆ E.g. An ideal resistor ˆ E.g. Aircraft: Mass (M) of aircraft

V (t) V (t+δ)
I(t) = R
⇒ I(t + δ) = R
changes as fuel is consumed.
F (t)
Acceleration : a(t) = M (t)

1.2.4 Causal Vs Non-causal Systems

Causal Systems Non-causal Systems

ˆ Output is only dependent on inputs ˆ Output depends on future inputs as well

already received (present or past)
ˆ System anticipates future inputs based
ˆ Non-anticipatory system on past

ˆ y(t) = f (x(t), x(t − 1), ...)) ˆ y(t) = f (x(t), x(t + 1), ...))

ˆ E.g. –Thermostat based AC

–Motor or generator ˆ E.g.–Weather forecasting system
–Missile guidance system

4
 Chapter 1. Modeling in Frequency Domain

1.3 What is a Control System?

1.3.1 Control System

ˆ A system or mechanism which directs the input to other systems and regulates
their output

ˆ Control system alters the response of a plant or system as desired

Desired Control Output

P lant or
Controller
Ref erence Input System

Control System T o be controlled

Figure 1.3: Control System

1.3.2 Control System with disturbance

ˆ Unwanted signals which affect the output of the system

ˆ E.g. People entering and leaving an AC room disturbs room temperature

ˆ Controller has to eliminate the effects of disturbance

Disturbance

Desired Control Output

P lant or
Controller
Ref erence Input System

Control System T o be controlled

Figure 1.4: Control System with disturbance

5
 Chapter 1. Modeling in Frequency Domain

1.3.3 Feedback in Control

ˆ Feedback senses the plant output and gives a signal which can be compared
to the reference

ˆ Controller action (control input) changes based on feedback

ˆ Feedback enables the control system in extracting the desired performance

from the plant even in presence of disturbance

Disturbance

Desired Control Output

P lant or
Controller
Ref erence Input System

Control System T o be controlled

F eedback

Figure 1.5: Feedback Control System with disturbance

1.4 Examples of Control Systems

1.4.1 Open-Loop Control system (Example–1)

Figure 1.6: Open-Loop Control system

6
 Chapter 1. Modeling in Frequency Domain

1.4.2 Open-Loop Control system (Example–2)

Figure 1.7: Open-Loop Control system

1.4.3 Closed-Loop (feedback) Control system (Example–1)

Figure 1.8: Closed-Loop Control system

1.4.4 Closed-Loop (feedback) Control system (Example–2)

Figure 1.9: Closed-Loop Control system

7
 Chapter 1. Modeling in Frequency Domain

1.5 Terminologies used in Control Systems

1.5.1 Elements of Control Systems

ˆ Plant : Room

ˆ Control System : Air conditioner

ˆ Reference : Desired temperature

ˆ Control Input : Compressor On/Off

ˆ Output : Output temperature

ˆ Disturbance : Factors affecting ambient temperature

ˆ Feedback : Measure temperature

P eople in the Room

Actual
T emp On / Of f Air temp
Desired
Setting Room
T emperature Switch Condition
Knob

Figure 1.10: Closed-Loop Control system

1.6 The Roadmap

1.6.1 The control system design process

Figure 1.11: The control system design process

8
 Chapter 1. Modeling in Frequency Domain

1.7 Modelling of Systems

1.7.1 What is a Model?

ˆ A model is an abstraction of the physical world.

ˆ Used for analysis and design, possibly before physical system exists.

ˆ Captures the dynamics of a system.

ˆ Dynamics refers to evolution of system variables.

– The room temperature when AC is ON

– The speed of car when accelerator is pushed by certain angle

– Current changing through inductor when voltage is applied

P lant or Equivalent
System M athematical
M odel

1.8 Model Derivation

1.8.1 First Principles (Analytical modelling)

ˆ From first principles (Analytical modelling)

– Use physical laws to derive models

– Combination of physical modelling and mathematical modelling

– Provides understanding

– Can use empirical data to determine parameters, validate model

P lant or Equivalent
System M athematical
M odel

9
 Chapter 1. Modeling in Frequency Domain

J θ̈ + bθ̇ = Ki
di
L dt + Ri = V − K θ̇

Figure 1.12: From first principles (Analytical modelling)–Example

1.8.2 Empirical Data (Experimental modelling)

ˆ From empirical data (Experimental modelling)

– Feed a known input and observe output, fit model to data

System

– Good for complicated systems (IC engine, battery)

– Good for black-box systems

– Does not provide intuition, can’t be widely applied

– Also called modelling by synthesis

Figure 1.13: Empirical Data (Experimental modelling–Example

10
 Chapter 1. Modeling in Frequency Domain

1.9 Types of Mathematical Models

Table 1.1: Types of Mathematical Models

Model Type Model Description

Differential Equation Model ˆ Dynamics of systems represented in terms of
differential equations

ˆ Time domain representation of the system

Transfer Function Model ˆ Dynamics represented in terms of a Laplace

transform expression

ˆ Frequency domain representation of the system

State Space Model ˆ State is a set of variables that describes the system
behaviour in conjunction with the system inputs

ˆ Dynamics are represented by a set of first order

differential equations using these state variables

1.9.1 Examples of Modelling

Differential Equation Model

dV dI d2 I I
=R +L 2 +
dt dt dt C
Transfer Function Model
I(s) 1
= 1
V (s) R + Ls + Cs

State Space Model

      
dIL
dt
−R
L
1
L
IL 1
L
 =  +  V
      
 
dVc 1
dt C
0 Vc 0

11
 Chapter 1. Modeling in Frequency Domain

1.10 Modelling Steps

1.10.1 Steps of Analytical Modelling

P urpose of the Def ine P ostulate a

model boundaries structure

M ath description
Select variable Apply relevant
of each model
of interest physical laws
element

F inal f orm of Analyse and M odif y model

mathematical
validate model if necessary
model

Figure 1.14: Modelling steps

1. Purpose of the model: Model to be developed should be based on specific

objective

2. Define boundaries: System of interest is separated from the rest of the

environment by a boundary

3. Postulate a structure: Represent the actual system as interconnection of

system elements

4. Select variables of interest: Assign variables to all system attributes of

interest

5. Mathematical description of each model elements: Identify the relation

between variables at each of the model elements

12
 Chapter 1. Modeling in Frequency Domain

6. Apply relevant physical laws: Develop equations to describe the effects

of element interconnections using physical laws

7. Final form of mathematical model: All the equations resulting from

step 6 put together to form the mathematical model of the system

8. Analyse and validate model: Verify the accuracy of the model

9. Modify model if necessary: To be done if result in step 8 are not convincing

1.11 Elements of Modelling

1.11.1 Physical Systems

ˆ Physical systems can be classified into various types:

X Electrical systems

X Mechanical systems

– Electronic systems

– Hydraulic systems

– Thermal systems

ˆ Each of these systems can be modelled in terms of certain basic elements

ˆ Basic elements of all physical systems can be shown to be analogous

13
 Chapter 1. Modeling in Frequency Domain

1.12 Electrical Systems

1.12.1 Types of sources

Based on the type of source, electrical systems can be classified as :

– Voltage sourced systems
– Current sourced systems

Voltage sourced Current sourced

Basic System Elements
Resistor (R) Resistor (R)
Inductor (L) Inductor (L)
Capacitor (C) Capacitor (C)

Basic System Variables

Voltage (V) Current (I)
Charge (Q) Flux (φ)

1.12.2 Types of sources

ˆ Voltage(e) – is measure of the force that causes electrons to move through a

circuit (a potential measured w.r.t. a ground)

ˆ Current(i) – is a measure of the rate of flow of charge (electrons) through a

dq
circuit (i = dt
), current has direction

14
 Chapter 1. Modeling in Frequency Domain

1.12.3 Elements of Electrical Systems

Electrical systems consist of three basic types of elements

1. Resistance elements

2. Capacitance elements

3. Inductance elements

1.12.4 Elements of Electrical Systems

ˆ Resistor(R): It is an element which resists the flow of current in an electrical

system.

V = IR (Dissipates Energy)

ˆ Capacitor(C):It is an element that stores electrical energy in a electric field

– Capacitance is measured as charge

stored per unit voltage

– If voltage is applied across a

q 1 1
R capacitor a potential builds up that
C= v
⇒v= C
q ⇒v= C
idt
is then released if the voltage is
removed... in other words, capacitors
store potential energy (like a spring)

ˆ Inductor(L):It is an element that stores electrical energy in a magnetic field

– Current through coil generates a magnetic field

which induces a voltage that is proportional to how
fast the current is changing

dφ
V = dt
⇒ V = L dI
dt
– If power is disconnected, induced voltage will make
the current continue to flow (like an inertia)

– Inductance elements store kinetic energy

15
 Chapter 1. Modeling in Frequency Domain

1.13 Analysis of Electrical Systems

Review: Steps of Modelling

P urpose of the Def ine P ostulate a

model boundaries structure

M ath description
Select variable Apply relevant
of each model
of interest physical laws
element

F inal f orm of Analyse and M odif y model

mathematical
validate model if necessary
model

1.13.1 Nodal and Loop Analysis

ˆ Nodal and loop analysis form part of Step 6 in the steps of modelling.

ˆ A structured way of applying physical laws and getting model equations.

ˆ Applied after identifying the basic elements of the system and variables of
interest.
N ode 1

Element Element Element

Loop 1 Loop 2
1 2 3

N ode 2

1. Nodal analysis based on Kirchhoff’s current law

2. Loop or mesh analysis based on Kirchhoff’s voltage law

16
 Chapter 1. Modeling in Frequency Domain

N ode 1 i3
i1
− +
i2

Element Element Element

V1 V2
1 2 3

+ Loop 1 − Loop 2

N ode 2

Kirchhoff ’s Current law (KCL) Kirchhoff ’s Voltage law (KVL)

At any node in an electrical circuit, the In an electrical circuit, the directed sum of
directed sum of currents flowing out of that voltages around a closed loop is zero.
node is equal to zero.

1.13.2 Nodal Analysis–Example

V N ode 1

+ + +

− − −

1.13.3 Mesh Analysis–Example

17
 Chapter 1. Modeling in Frequency Domain

1.14 Mechanical Systems

1.14.1 Classification
Mechanical systems can be classified into two types:
–Translational systems having linear motion
–Rotational systems having angular motion about a fixed axis
Translational Rotational

Basic System Elements

Mass (M) Inertia (J)

Damper (B) Damper (D)

Linear spring (K) Torsional spring (K)

Basic System Variables

Force (F) Torque (T)

Displacement (x) Angular displacement (θ)

1.14.2 Elements of Translational Systems

ˆ Spring(K): Spring is an element, which stores potential energy.

– If a force is applied on spring K, then it is opposed

by an opposing force due to elasticity of spring.

– This opposing force (fk ) is proportional to the

displacement (x) of the spring.

fk ∝ x
fk = Kx

By Newton’s second law

F = fk
F = Kx

18
 Chapter 1. Modeling in Frequency Domain

ˆ Mass (M)): Mass is the property of a body, which stores kinetic energy.

– If a force is applied on a body having mass M, then

it is opposed by an opposing force (fm ) due to mass.

– This opposing force is proportional to the

acceleration (a) of the body.

fm ∝ a
fm = M a
d2 x
fm = M 2
dt
By Newton’s second law,

F = fm
d2 x
F = M
dt2

ˆ Damper (B):
– If a force is applied on dashpot B, then it is opposed
by an opposing force (fb ) due to friction of the
dashpot.

– This opposing force is proportional to the velocity

(v) of the body.

fb ∝ v
fb = Bv
dx
fb = B
dt
By Newton’s second law,

F = fb
dx
F = B
dt

19
 Chapter 1. Modeling in Frequency Domain

1.14.3 Units for Translational Systems

Table 1.2: Variables of mechanical translational system

Symbol Quantity Units

f (t) Force N ewtons(N )
x(t) Displacement meters(m)
v Velocity meters/sec(m/s)
a(t) Acceleration meters/sec2 (m/s2 )
M Mass kilograms(kg)
K Spring stiffness N ewtons/meter(N/m)
B Viscous-friction N ewton − seconds/meter(N − s/m)
coefficient

1.14.4 Example of Spring Mass Damper

Automobile Suspension system

20
 Chapter 1. Modeling in Frequency Domain

1.14.5 Freebody Diagram

ˆ To obtain mathematical model of mechanical systems, it is necessary to draw

free-body diagram indicating various forces acting on it.

ˆ In a free-body diagram each mass is considered separately with all forces

acting on it.

Procedure to draw free-body diagram

ˆ Assume the direction of displacement of the given mass as positive direction.

ˆ Find all the forces and its direction acting on the mass.

ˆ E.g. Force applied is in the direction of displacement and the opposing forces
due to spring, mass and damper are opposite to the direction of displacement.

ˆ Using Newton’s laws of motion, express all the forces in terms of displacement
or velocity of the mass.

1.14.6 Free-body diagram–Example

Original System Free-body diagram

21
 Chapter 1. Modeling in Frequency Domain

1.15 Electrical Analogies of Mechanical Systems

1.15.1 Force Voltage Analogy

Translational System Electrical System

Force (F) Voltage (V)

Mass (M) Inductance (L)

Frictional Coefficient (B) Resistance (R)

Spring Constant (K) Reciprocal of Capacitance ( C1 )

Displacement (x) Charge (q)

Velocity(v) Current (i)

1.15.2 Force Current Analogy

Translational System Electrical System

Force (F) Current (i)

Mass (M) Capacitance (C)

Frictional Coefficient (B) Reciprocal of Resistance ( R1 )

Spring Constant (K) Reciprocal of Inductance ( L1 )

Displacement (x) Magnetic Flux (φ)

Velocity(v) Voltage (V)

22
 Chapter 1. Modeling in Frequency Domain

1.16 Transfer Function

1.16.1 Need of Transfer Function

Motivation

ˆ To find the time response, we need to solve ordinary differential equations

(integro-differential equations).

ˆ When the model equations are transformed to s–domain, they turn out to
be algebraic equations which are easier to solve.

ˆ The transformed model in s–domain is called transfer function model.

ˆ It is a model which is applicable for all kinds of input signals.

1.16.2 What is a Transfer Function?

ˆ For an LTI system, transfer function is the ratio of the Laplace transform of
the output to the Laplace transform of the input with the initial conditions
being zero.

ˆ Mathematically, if R(s) is the Laplace transform of the input function and

C(s) is the Laplace transform of the output, the transfer function G(s) is
given by:

C(s)
G(s) =
R(s)

1.16.3 Impulse response

ˆ Impulse signal (δ(t)) is infinitesimally narrow and

infinitely tall yet integrating to one.

ˆ It takes zero value everywhere except at t = 0

Z ∞
δ(t) = 1
−∞

23
 Chapter 1. Modeling in Frequency Domain

ˆ If input to the system is the unit impulse, then the

output is called the impulse response i.e.,
u(t) = δ(t) ⇒ U (s) = 1 ⇒ G(s) = C(s)

ˆ That means transfer function is the Laplace

transform of the impulse response of an LTI system
when the initial conditions are set to zero.

1.17 Transfer Function Steps

1.17.1 Steps to find Transfer function

1. Find the model equations of the given system.

2. Identify the system input and output variables.

3. Take the Laplace transform of the model equations assuming zero initial
conditions.

4. Find the ratio of the Laplace transform of the output to the Laplace transform
of the input.

Examples
Eo (s)
E.g.1 Obtain the transfer function of the electrical system shown in Figure 1.18.
Ei (s)

Figure 1.18: Electrical circuit

24
 Chapter 1. Modeling in Frequency Domain

Eo (s)
E.g.2 Obtain the transfer function of the electrical system shown in Figure 1.19.
Ei (s)

Figure 1.19: Electrical circuit

X(s)
E.g.3 Find the transfer function for the system of Figure 1.20.
F (s)

Figure 1.20: Mechanical system

Y2 (s)
E.g.4 Obtain the transfer function for system shown in Figure 1.21.
U (s)

Figure 1.21: Mechanical system

25
 Chapter 1. Modeling in Frequency Domain

1.17.2 Properties of Transfer Function

ˆ Transfer function of a system is independent of the magnitude and nature of

input.

ˆ Using the transfer function, the response can be studied for various inputs
to understand the nature of the system.

ˆ Transfer function does not provide any information concerning the physical
structure of the system i.e., two different physical systems can have the same
transfer function.

1.18 Block Diagram

1.18.1 Motivation

ˆ How to visualise a complex system with many components?

ˆ How to understand the flow and transformation of signals in a complex

system?

ˆ How to reduce complex system to a single block?

ˆ How to find the transfer function of a complex system?

Block Diagram Representation

1.18.2 Block Diagram of a System

ˆ It is a pictorial representation of the system which depicts

– Each function component or sub-system and

– Flow of signals from one sub-system to another

ˆ Block diagram provides a simple representation of complex systems.

ˆ Block diagram enables calculating the overall system transfer function provided
the transfer functions of each of the components or sub-systems are known.

26
 Chapter 1. Modeling in Frequency Domain

1.19 Components of Block Diagram

1.19.1 Basic Elements of Block Diagram

Block diagrams have four components:

Figure 1.22: Block diagram of a closed loop system

1.19.2 Basic Elements of Block Diagram

ˆ Block: The transfer function of a component is represented by a block. Block

has single input and single output.

G(s) is the transfer function of sub-system

ˆ Arrows: To represent the flow of signals.

ˆ Summing points: To represent summation of two or more signals.

27
 Chapter 1. Modeling in Frequency Domain

ˆ Take-off points: To represent the branching of a signal.

1.20 Developing Block Diagram from Mathematical

model

1.20.1 Example

1.21 Block Diagram Algebra

1.21.1 Basic connections for Blocks
ˆ Series connection:

– Components or sub-systems are connected in series each having its own

transfer function.

– Overall transfer function is product of individual transfer functions

28
 Chapter 1. Modeling in Frequency Domain

Figure 1.23: Cascaded subsystems

C(s)
G(s) = = G3 (s)G2 (s)G1 (s)
R(s)

Figure 1.24: Equivalent transfer function

ˆ Parallel Connection:

– Components or sub-systems of a system are connected in parallel.

– Overall transfer function is sum of individual transfer functions

Figure 1.25: Parallel subsystems

C(s)
G(s) = = G1 (s) ± G2 (s) ± G3 (s)
R(s)

Figure 1.26: Equivalent transfer function

29
 Chapter 1. Modeling in Frequency Domain

1.22 Block Diagram Reduction

ˆ Block diagram reduction refers to simplification of block diagrams of complex

systems through certain arrangements.

ˆ Simplification enables easy calculation of the overall transfer function of the

system.

ˆ Simplification is done using certain rules called the ‘rules of block diagram
algebra’.

ˆ All these rules are derived by simple algebraic manipulations of the equations
representing the blocks.

1.22.1 Block Diagram Algebra for Summing Points

Shifting summing point after the block

Figure 1.27: Moving a block to the left past a summing junction

Shifting summing point before the block

Figure 1.28: Moving a block to the right past a summing junction

30
 Chapter 1. Modeling in Frequency Domain

1.22.2 Block Diagram Algebra for Take-off Points

Shifting take-off point after the block

Figure 1.29: Moving a block to the left past a pickoff point

Shifting take-off point before the block

Figure 1.30: Moving a block to the right past a pickoff point

1.22.3 Block Diagram Reduction Rules

Rule 1 – Check for the blocks connected in series and simplify.

Rule 2 – Check for the blocks connected in parallel and simplify.

Rule 3 – Check for the blocks connected in feedback loop and simplify.

Rule 4 – If there is difficulty with take-off point while simplifying, shift it towards
right.

Rule 5 – If there is difficulty with summing point while simplifying, shift it

towards left.

Rule 6 – Repeat the above steps till you get the simplified form, i.e., single block.

31
 Chapter 1. Modeling in Frequency Domain

Examples
C(s)
E.g.1: Obtain the transfer function for the system shown in Figure 1.31.
R(s)

Figure 1.31: Block diagram of a system

C(s)
E.g.2: Obtain the transfer function for the system shown in Figure 1.32.
R(s)

Figure 1.32: Block diagram of a system

C(s)
E.g.3: Find the transfer function for the system of Figure 1.33.
R(s)

Figure 1.33: Block diagram of a system

32
 Chapter 1. Modeling in Frequency Domain

1.23 Signal Flow Graphs

1.23.1 What is Signal Flow Graph?

ˆ Signal flow graph is an alternate approach to block diagram representation.

ˆ It is a graphical representation of the relationships between the system variables.

ˆ Signal flow graph does not require any reduction because there is a graph
gain formula which directly gives the transfer function.

ˆ It was developed by S. J. Mason

ˆ For complicated systems, when block diagram reduction method becomes

tedious and time consuming then signal flow graph is a good choice.

1.23.2 The Difference

Block Diagram Signal Flow graph

In this method, at each step block diagram Only one time Signal flow graph is to be
is to be redrawn. Thats why it is tedious drawn and then apply Mason’s gain formula
method. to evaluate the system.

1.23.3 Properties of Signal Flow Graphs

ˆ Signal flow graph can be obtained from the block diagram of a system.

ˆ Signal flow graph can be obtained from system equations directly.

ˆ The equations must be in algebraic form (s – domain).

ˆ It is applicable only for linear systems.

33
 Chapter 1. Modeling in Frequency Domain

1.24 Basic Elements of Signal Flow Graph

Signal Flow Graph of a system

1.24.1 Basic Elements

1. Node: Node is a point which represents either a variable or a signal.

2. Branch: Branch is a line segment which joins two nodes. It has both gain
and direction.

3. Source or Input Node: A node which has only outgoing branches is known
as source or input node.
e.g. Node y1 is a source node.

4. Sink or Output Node: A node which has only incoming branches is known
as sink or output node.
e.g. Node y6 is a sink node.

5. Chain Node: A node having incoming and outgoing branches is known as

chain node.
e.g. Node y2 , y3 , y4 and y5

6. Forward path: A path from input node to output node is defined as forward
path.
e.g. y1 – y2 – y3 – y4 – y5 – y6 First forward path
y1 – y2 – y5 – y6 Second forward path

34
 Chapter 1. Modeling in Frequency Domain

7. Path gain: The product of branch gains is known as path gain.

e.g. Path gain for first forward path y1 – y2 – y3 – y4 – y5 – y6 is a1 a3 a4 a5 a7
Path gain for second forward path y1 – y2 – y5 – y6 is a1 a6 a7

8. Feedback loop: A loop consisting of more than one node but originating and
terminating at the same node is known as feedback loop.
e.g. y3 – y4 – y5 – y3 or y2 – y3 – y2 or y2 – y5 – y3 – y2

9. Self loop: A feedback loop containing only one node is called as self loop.
e.g. y2 – y2 having gain a2

10. Non–touching loops: While forming different loops, if the same node is
not traced again i.e. if the node is not common, then those are called as
non–touching loops.

11. Loop gain: The product of all the branch gains forming a loop is known as
loop gain.

12. Dummy node: Separate input and output nodes added with branch gain 1
are known as dummy nodes.

35
 Chapter 1. Modeling in Frequency Domain

1.25 Procedure to draw signal flow graph

1.25.1 Signal flow graph from Block diagram
1. Step 1: Replace the input signal and output signal by nodes.

2. Step 2: Replace all the summing points by nodes.

3. Step 3: Replace all the take-off points by nodes.

4. Step 4: If the branch connecting a summing point and take-off point has
unity gain, then the summing point and take-off point can be combined and
represented by a single node.

5. Step 5: If there are more than one take-off points from the same signal then
all the take-off points can be combined and represented by a single node.

6. Step 6: If the gain of the link connecting two summing points is one, then
the two summing points can be combined and can be replaced by a single
node.

7. Step 7: In summing point, if a signal is subtracted instead of addition then

multiply the transmittance by -1 while representing in signal flow graph.

1.25.2 Mason’s Gain formula

Purpose of Mason’s gain formula is to find the overall gain of the signal flow graph
which is the transfer function of the system. Suppose there are ‘N’ forward paths
in a signal flow graph. It can be calculated by using
PN
C(s) Pi ∆i
G(s) = = i=1
R(s) ∆
where,
C(s) is the output node
R(s) is the input node
G(s) is the transfer function or gain between R(s) and C(s)
Pi is the ith forward path gain
∆ = 1 – (sum of all individual loop gains) + (sum of gain products of all possible two
non-touching loops) – (sum of gain products of all possible three non-touching loops)+...
∆i is obtained from ∆ by removing the loops which are touching
the ith forward path.

36
 Chapter 1. Modeling in Frequency Domain

Examples
C(s)
E.g.1: Using Mason’s rule, find the transfer function, T (s) = for the system
R(s)
represented in Figure 1.35.

Figure 1.35: Signal flow graph of a system

C(s)
E.g.2: Using Mason’s rule, find the transfer function, T (s) = for the system
R(s)
represented in Figure 1.36.

Figure 1.36: Signal flow graph of a system

E.g.3: For the linear, time-invariant system whose block diagram is shown in
Y (s)
Figure 1.37 find the transfer function .
X(s)

Figure 1.37: Signal flow graph of a system

37
 Chapter 1. Modeling in Frequency Domain

E.g.4: For the linear, time-invariant system whose signal flow graph is shown in
Y (s)
Figure 1.38 find the transfer function .
X(s)

Figure 1.38: Signal flow graph of a system

E.g.5: The block diagram of a system is shown in the Figure 1.39. If the desired
s
transfer function of the system is C(s)
R(s)
= 2 , then G(s) is:
s +s+2
s
(a) 2 (b) s (c) 1/s (d) 1
s +s+2

Figure 1.39: Block diagram of a system

E.g.6: Which of the options is an equivalent representation of the signal flow

graph shown here?

(a) (b)

(c) (d)

38
 Chapter 1. Modeling in Frequency Domain

E.g.7: The block diagram for a particular control system is shown in the figure.
C(s)
What is the transfer function R(s)
for this system?

(a) (b)
s+a s−b
s−b s+a
(c) (d)
s+a s+b
s+b s+a
Vo (s)
E.g.8: Find the transfer function Vi (s)
for the lag network shown in Figure 1.40.

Figure 1.40: Lag network

Vo (s)
E.g.9: Find the transfer function Vi (s)
for the lead network shown in Figure 1.41.

Figure 1.41: Lead network

39
 Chapter 1. Modeling in Frequency Domain

X1 (s) X2 (s)
E.g.10: Obtain the transfer function U (s)
and U (s)
of the mechanical system
shown in Figure 1.42.

Figure 1.42: Mechanical system

X1 (s)
E.g.11: Find the transfer function, G(s) = F (s)
, for the translational mechanical
system shown in Figure 1.43.

Figure 1.43: Mechanical system

X2 (s)
E.g.12: Find the transfer function, G(s) = F (s)
, for the translational mechanical
network shown in Figure 1.44.

Figure 1.44: Mechanical system

40
Chapter 2

Modeling in Time Domain

Learning Outcomes
After completing this chapter, students will be able to:

ˆ Explain importance of state space

ˆ Demonstrate different state models for SISO and MIMO systems

ˆ Construct Transfer Function from state model

ˆ Determine time response from state transition matrix

ˆ Examine importance of controllability and observability in the view of stability

41
 Chapter 2. Modeling in Time Domain

2.1 Introduction to State Space systems

2.1.1 Transfer function modelling approach

The good stuff! The drawbacks

ˆ Transfer functions are defined in ˆ Can be used for only LTI systems
the Laplace domain
– Can be extended for (some) nonlinear

ˆ System: input–output view systems, but tricky!

– Input can be signals of various frequencies ˆ Limited to single input single

– Bandwidth of the control system

output systems

ˆ Algebraic properties of the transfer – Can handle MIMO systems, with some

jugglery!
functions

– Combine different systems

ˆ Internal behaviour of the system is
not known – black box!
ˆ Very dominant in the industry!
ˆ Gives only zero state response

2.1.2 Advantages of state space method

ˆ It is valid for linear and non-linear systems that have backlash, saturation
and dead zone.

ˆ State space approach can handle the systems with non-zero initial conditions.

ˆ State space method works directly with the governing differential equations
in the time domain.

ˆ State space approach is applicable to time varying systems.

ˆ State space equation is applicable to single input single output as well as

multi-input multi-output systems.

ˆ State equations are always first order irrespective of the system’s order, input
and outputs.

42
 Chapter 2. Modeling in Time Domain

2.2 Definitions
2.2.1 Definitions related to state space techniques
State The concept of state is related to minimum set of variables. These variables
are called as state variable. Thus, the state of a system is defined as the
minimum (smallest) set of variables such that knowledge of these variables
at t = t0 , along with the knowledge of the input for t ≥ t0 , completely
determines the behaviour of the system for any time.

State vector If n state variables represents behaviour of a given system, then n

state variables can be considered as n components of a state vector x. Such
vector is called as state vector.

State space The n dimensional space whose coordinate axes consist of the x1
axis, x2 axis,..., xn axis is called state space.

State equations It is the set of n simultaneous first order differential equation

with n variables.

Output equations It is the linear combinations of the state variables and the
inputs.
State variables: [x1 (t), x2 (t),...,xn (t)]

u1 (t) y1 (t)
nth order

um (t) system yp (t)

System block diagram

u1 (t) y1 (t)
Input Controlled System Output
Variables u2 (t) y2 (t) Variables
m p
State variables (n)
um (t) yp (t)

x1 x2 xn

MIMO system

43
 Chapter 2. Modeling in Time Domain

2.3 Mathematical Representation of state space

2.3.1 State equation

The state representation can be arranged in the form of n first order equations
dx1
= ẋ1 (t) = f1 (x1 , x2 , ..., xn ; u1 , u2 , ...um ; t)
dt
dx2
= ẋ2 (t) = f2 (x1 , x2 , ..., xn ; u1 , u2 , ...um ; t)
dt
..
.
dxn
= ẋn (t) = fn (x1 , x2 , ..., xn ; u1 , u2 , ...um ; t)
dt
ẋ(t) = f (x, u, t)

2.3.2 Output equation

y1 (t) = g1 (x1 , x2 , ..., xn ; u1 , u2 , ...um ; t)

y2 (t) = g2 (x1 , x2 , ..., xn ; u1 , u2 , ...um ; t)
..
.
yp (t) = gn (x1 , x2 , ..., xn ; u1 , u2 , ...um ; t)

y(t) = g(x, u, t)

2.4 State space equations – Matrix form

2.4.1 State equation – Matrix form
State model of a linear time invariant system is a special case of the general time
invariants models:
In this case, each state variable now becomes linear combination of system states
and inputs, i.e.,

44
 Chapter 2. Modeling in Time Domain

ẋ1 (t) = a11 x1 + a12 x2 + ... + a1n xn + b11 u1 + b12 u2 + ... + b1m um
ẋ2 (t) = a21 x1 + a22 x2 + ... + a2n xn + b21 u1 + b22 u2 + ... + b2m um
..
.
ẋn (t) = an1 x1 + an2 x2 + ... + ann xn + bn1 u1 + bn2 u2 + ... + bnm um

In vector matrix form,

ẋ(t) = Ax(t) + Bu(t)

2.4.2 Output equation – Matrix form

Output variables at time ‘t’ are linear combination of the values of the input and
state variables at time ‘t’, i.e.,

y1 (t) = c11 x1 + c12 x2 + ... + c1n xn + d11 u1 + d12 u2 + ... + d1m um

y2 (t) = c21 x1 + c22 x2 + ... + c2n xn + d21 u1 + d22 u2 + ... + d2m um
..
.
yp (t) = cp1 x1 + cp2 x2 + ... + cpn xn + dp1 u1 + dp2 u2 + ... + dpm um

In vector matrix form,

y(t) = Cx(t) + Du(t)

2.4.3 State space Matrices

Differential state vector: It is a collection of n differential state variables

 
 ẋ1 
 
 ẋ2 
[ẋ]n×1 =  . 
 
 .. 
 
 
ẋn
n×1

45
 Chapter 2. Modeling in Time Domain

State vector: It is a collection of n state variables

 
 x1 
 
 x2 
[x]n×1 =  . 
 
 .. 
 
 
xn
n×1

Input vector: It is a collection of m input variables

 
 u1 
 
 u2 
[u]m×1 =  . 
 
 .. 
 
 
um
m×1

System matrix: It is a square matrix

 
 a11 a12 . . . a1n 
 
 a21 a22 . . . a2n 
[A]n×n =  .
 

 .. 
 
 
an1 an2 . . . ann
n×n

Input matrix: It is a input matrix which relates input and state variables.
 
 b11 b12 . . . b1m 
 
 b21 b22 . . . b2m 
[B]n×m =  .
 

 .. 
 
 
bn1 bn2 . . . bnm
n×m

Output vector: It is a collection of p output variables

 
 y1 
 
 y2 
[y]p×1 =  . 
 
 .. 
 
 
yp
p×1

46
 Chapter 2. Modeling in Time Domain

Output matrix: It is a output matrix which relates output and state variables.
 
c11 c12 . . . c1n 
 
c21 c22 . . . c2n 
[C]p×n =  .
 

 .. 
 
 
cp1 cp2 . . . cpn
p×n

Data transmission matrix: It is a matrix which relates input and output.

 
d11 d12 . . . d1m 
 
d21 d22 . . . d2m 
[D]p×m =  .
 

 .. 
 
 
dp1 dp2 . . . dpm
p×m

2.4.4 Block diagram representation of state space model

ẋ = Ax + Bu
yẋ =
= Cx
Ax +
+ Du
Bu
y = Cx + Du

U
B +
ẋ R x C +
Y

U
B +
+
ẋ R x C +
+ Y

+ +

U
B +
ẋ R x C +
Y

+ A
+

Block diagram representation of state space

47
 Chapter 2. Modeling in Time Domain

2.4.5 Example

ẋ = Ax + Bu
y = Cx + Du

U
 B  + + ẋ R
 x  C  +
Y

+
dVC 1
dt
0 C
VC 0
 =  +  Vs
      
 
dIL −1 −R 1
dt L A
L
IL L

D
 
h i h i VC
= +0
 
Y 1 0 
IL

2.5 Selection of state variables

1. Number of state variables should be minimum.

2. Number of state variables should be linearly independent.

3. Number of state variables is equal to number of independent energy storage

elements.

4. Number of state variables is equal to the order of differential equation.

5. Number of state variables are equal to number of integrators.

48
 Chapter 2. Modeling in Time Domain

2.6 Companion form (controllable canonical form)

2.6.1 Introduction
ˆ The companion form of modelling is called phase variable form.

ˆ They can be easily determined if the system model is already known in the
differential equation/ transfer function form.

ˆ They are defined as particular state variables which are obtained from one
of system variables and its derivatives.

ˆ Often variable is selected as the system output and the remaining variables
are the derivative of the output.

2.6.2 State space representation in controllable canonical

form – Case 1

Consider the nth order dynamical system,

dn y dn−1 y dy dm u dm−1 u du
+ a n−1 + . . . + a 1 + a 0 y = b m + b m−1 + . . . + b 1 + b0 u
dtn dtn−1 dt dtm dtm−1 dt
Case 1: Let the derivatives of u to be zero
dn y dn−1 y dy
+ a n−1 + . . . + a 1 + a0 y = b 0 u (2.1)
dtn dtn−1 dt
where, u is input and y is the output
Strategy: Select output as a first state variable (i.e. y) and remaining state
variables are derivatives of the output.

x1 = y
dy
x2 = = ẋ1
dt
d2 y
x3 = 2 = ẋ2
dt
..
.
dn−1 y
xn = = ẋn−1
dtn−1
dn y
ẋn = n (2.2)
dt

49
 Chapter 2. Modeling in Time Domain

∴ From equation (2.1) we can write ẋn as,

dn y dn−1 y dn−2 y dy
ẋn = n
= −a n−1 n−1
− a n−2 n−2
− . . . − a1 − a0 y + b 0 u (2.3)
dt dt dt dt
Substituting equation (2.2) in equation (2.3) we get,
dn y
ẋn = = −an−1 xn − an−2 xn−1 − . . . − a1 x2 − a0 x1 + b0 u (2.4)
dtn
Finally, from equation (2.2) and equation (2.4) we get the state variables as,

ẋ1 = x2
ẋ2 = x3
ẋ3 = x4
..
.
ẋn = −an−1 xn − an−2 xn−1 − . . . − a1 x2 − a0 x1 + b0 u (2.5)

State equation:
      
ẋ 0 1 0 ... 0 x1 0
 1      
      
      
 ẋ2   0 0 1 ... 0  x2   0 
 = + u
      
 ..  .. .. ..
  
    
 .   .  .   . 
      
      
ẋn −a0 −a1 −a2 . . . −an−1 xn b0

Output equation:
 
x1
 
 
 
h i h i x2 
=
 
y 1 0 ... 0 
..

 

 . 

 
xn

50
 Chapter 2. Modeling in Time Domain

Examples – Case 1

E.g.1: Find the state space model for the given transfer function
Y (s) 2
= 2
U (s) s +s+1

E.g.2: Find the state space model for the given transfer function
Y (s) 10
= 4
U (s) s + 3s + 2s2 + 5s + 4
3

E.g.3: Find the state space model for the differential equation given as
...
y + 3 ÿ + 2 ẏ = u

2.6.3 Controllable canonical form – Case 2

Consider the nth order dynamical system,

dn y dn−1 y dy dm u dm−1 u du
+ a n−1 + . . . + a 1 + a 0 y = b m + b m−1 + . . . + b 1 + b0 u
dtn dtn−1 dt dtm dtm−1 dt
Case 2: Consider the derivatives of u and convert to s domain

Y (s) bm sm + bm−1 sm−1 + . . . + b1 s + b0 B

G(s) = = n n−1
= (2.6)
U (s) s + an−1 s + . . . + a1 s + a0 A
Equation (2.6) can be split in two parts as,

U(s) 1 X(s) Y(s)

B
A

Y (s) X(s) Y (s)

= ×
U (s) U (s) X(s)
Let n = m = 3,
Y (s) b3 s 3 + b2 s 2 + b1 s + b0 B
G(s) = = 3 2
= (2.7)
U (s) s + a2 s + a1 s + a0 A
From equation (2.7) we can write,
X(s) 1 1 Y (s)
= = 3 2
and = B = b3 s 3 + b2 s 2 + b1 s + b0
U (s) A s + a2 s + a1 s + a0 X(s)

51
 Chapter 2. Modeling in Time Domain

X(s)A = U (s) ⇒ s3 X(s) + a2 s2 X(s) + a1 sX(s) + a0 X(s) = U (s) (2.8)

Converting equation (2.8) to time domain,

...
x (t) + a2 ẍ(t) + a1 ẋ(t) + a0 x(t) = u(t) (2.9)

Strategy: From equation (2.9) select output as state variable

x1 = x
dx
x2 = = ẋ1
dt
d2 x
x3 = 2 = ẋ2
dt
...
ẋ3 = x (t) = −a2 ẍ(t) − a1 ẋ(t) − a0 x(t) + u(t) (2.10)

From equation (2.10) we can write,

ẋ1 = x2
ẋ2 = x3
...
ẋ3 = x (t) = −a2 x3 − a1 x2 − a0 x1 + u(t) (2.11)

Now let us consider,

Y (s) = BX(s) ⇒ b3 s3 X(s) + b2 s2 X(s) + b1 sX(s) + b0 X(s) (2.12)

Converting equation (2.12) to time domain,

...
y(t) = b3 x (t) + b2 ẍ(t) + b1 ẋ(t) + b0 x(t) (2.13)

Substituting equation (2.11) in equation (2.13) we get,

y(t) = b3 (−a2 x3 − a1 x2 − a0 x1 + u(t)) + b2 x3 + b1 x2 + b0 x1 (2.14)

Rearranging equation (2.14) we get,

y(t) = (b0 − b3 a0 )x1 + (b1 − b3 a1 )x2 + (b2 − b3 a2 )x3 + b3 u(t) (2.15)

From equation (2.11) & equation (2.15) we get

State equation
      
ẋ1 0 1 0 x1 0
      
      
 ẋ2  =  0   x2  +  0  u
      
0 1
      
      
ẋ3 −a0 −a1 −a2 x3 1

52
 Chapter 2. Modeling in Time Domain

Output equation
 
x
 1 
h i h i 
= (b0 − b3 a0 ) (b1 − b3 a1 ) (b2 − b3 a2 )  x2  + b3 u
 
y
 
 
x3

For strictly proper transfer function (m < n) i.e. m = 2, n = 3 (b3 = 0) we get,

State equation
      
ẋ1 0 1 0 x1 0
      
      
 ẋ2  =  0   x2  +  0  u
      
0 1
      
      
ẋ3 −a0 −a1 −a2 x3 1

Output equation
 
x
 1 
h i h i  
=
 
y b0 b1 b2  x 2 
 
 
x3

Examples – Case 2

E.g.1: Find the state space model for the given transfer function

Y (s) 2s3 + s2 + 7s + 2
= 3
U (s) s + 15s2 + 62s + 48

E.g.2: Find the state space model for the given transfer function

Y (s) s4 + 2s3 + 12s2 + 7s + 6

= 5
U (s) s + 9s4 + 13s3 + 8s2

E.g.3: Find the state space model for the differential equation given as
Y (s) 8s + 10
= 4
U (s) 8s + 5s + s2 + 5s + 13
3

53
 Chapter 2. Modeling in Time Domain

E.g.4: For the given circuit, which one of the following is the correct state
equation?

(a) (b)

(c) (d)

2.7 Introduction to observable canonical form

2.7.1 Introduction

ˆ Observable canonical form is also useful in analyzing and designing control

systems because this form guarantees observability.

ˆ A system is observable if all its states can be determined by the output.

ˆ Observability is useful because it means the initial condition of a system can

be back calculated from what can be physically measured.

54
 Chapter 2. Modeling in Time Domain

2.7.2 Mathematical representation of observable canonical

form

Consider the nth order dynamical system,

dn y dn−1 y dy dm u dm−1 u du
+ a n−1 + . . . + a 1 + a 0 y = b m + b m−1 + . . . + b 1 + b0 u
dtn dtn−1 dt dtm dtm−1 dt
Above equation can be written in s domain as,

Y (s) bm sm + bm−1 sm−1 + . . . + b1 s + b0

G(s) = = (2.16)
U (s) sn + an−1 sn−1 + . . . + a1 s + a0
Let n = m = 3,

Y (s) b3 s 3 + b2 s 2 + b1 s + b 0
G(s) = = 3 (2.17)
U (s) s + a2 s 2 + a1 s + a0
Equation (2.17) can be written in time domain form as,

d3 y d2 y dy d3 u d2 u du
+ a 2 + a 1 + a 0 y = b 3 + b 2 + b 1 + b0 u (2.18)
dt3 dt2 dt dt3 dt2 dt
Rearranging (2.18) we get,

d3 y d3 u d2 d
= b 3 + (b 2 u − a 2 y) + (b1 u − a1 y) + b0 u − a0 y (2.19)
dt3 dt3 dt2 dt
Integrating equation (2.19) we get,

Z Z Z Z Z Z
y = b3 u + (b2 u − a2 y)dt + (b1 u − a1 y)dt + (b0 u − a0 y)dt (2.20)

Drawing block diagram from equation (2.20) as shown below,

u

b0 b1 b2 b3

ẋ1 R x1 ẋ2 R x2 ẋ3 R x3 y

−a0 −a1 −a2

Figure 2.4: Block diagram of equation (2.20)

55
 Chapter 2. Modeling in Time Domain

From the figure (2.4) we can write,

ẋ1 = b0 u − a0 y
= b0 u − a0 (x3 + b3 u)
ẋ1 = −a0 x3 + u(b0 − a0 b3 ) (2.21)

ẋ2 = x1 + b1 u − a1 (x3 + b3 u)
= x1 + b1 u − a1 x3 − a1 b3 u)
ẋ2 = x1 − a1 x3 + u(b1 − a1 b3 ) (2.22)

ẋ3 = x2 + b2 u − a2 (x3 + b3 u)
= x 2 − a2 x 3 + b 2 u − a2 b 3 u
ẋ3 = x2 − a2 x3 + u(b2 − a2 b3 ) (2.23)

From equation (2.21),(2.22) and (2.23) we can write state equation as,
State equation
      
ẋ 0 0 −a0 x (b − a0 b3 )
 1    1   0 
      
 =  1 0 −a1   x2  +  (b1 − a1 b3 ) u
      
 ẋ2
      
      
ẋ3 0 1 −a2 x3 (b2 − a2 b3 )

From the block diagram, we know that output y = x3 + b3 u

Output equation
 
x1
 
h i h i 
= x  + b3 u
 
y 0 0 1
 2 

 
x3

56
 Chapter 2. Modeling in Time Domain

Note the relationship between the observable and controllable forms:

Aobs = ATcont
T
Bobs = Ccont
T
Cobs = Bcont
Dobs = Dcont

Examples

E.g.1: Find the state space model for the given transfer function
Y (s) 2s3 + s2 + 7s + 2
= 3
U (s) s + 15s2 + 62s + 48

E.g.2: Find the state space model for the given transfer function
Y (s) s4 + 2s3 + 12s2 + 7s + 6
= 5
U (s) s + 9s4 + 13s3 + 8s2

E.g.3: Find the state space model for the differential equation given as
Y (s) 8s + 10
= 4
U (s) 8s + 5s + s2 + 5s + 13
3

2.8 Introduction to diagonal canonical form

ˆ The state models having minimum number of non-zero elements are called
as canonical forms.

ˆ There are two types: (i) Diagonal canonical form (ii) Jordan canonical form.

ˆ The diagonal canonical form is a state space model in which the poles of the
transfer function are arranged diagonally in the A matrix.

ˆ This method of state space modelling is based on the concept of partial

fraction expansion.

ˆ This form is useful in checking controllability, observability, state transition

matrix and also in model order reduction.

57
 Chapter 2. Modeling in Time Domain

2.9 Mathematical representation of diagonal canonical

form

2.9.1 Diagonal canonical form – Case 1

Consider the nth order dynamical system,

dn y dn−1 y dy dm u dm−1 u du
an + a n−1 + . . . + a 1 + a 0 y = b m + b m−1 + . . . + b 1 + b0 u
dtn dtn−1 dt dtm dtm−1 dt
Above equation can be written in s domain as,

Y (s) bm sm + bm−1 sm−1 + . . . + b1 s + b0

G(s) = = (2.24)
U (s) an sn + an−1 sn−1 + . . . + a1 s + a0

Case 1: Strictly proper transfer function (m < n)

Y (s) b 2 s 2 + b1 s + b0
G(s) = = (2.25)
U (s) a3 s 3 + a2 s 2 + a1 s + a0

The system transfer function (2.25) having a denominator polynomial that can

be factored into distinct (λ1 6= λ2 6= . . . 6= λn ) roots as follows:

Y (s) c1 c2 c3
= + + (2.26)
U (s) (s + λ1 ) (s + λ2 ) (s + λ3 )

Re-arranging (2.26) we get (roots of denominator are also known as eigen values),

Y (s) c1 (s + λ2 )(s + λ3 ) + c2 (s + λ1 )(s + λ3 ) + c3 (s + λ1 )(s + λ2 )

= (2.27)
U (s) (s + λ1 )(s + λ2 )(s + λ3 )

Comparing equation (2.25) and equation (2.27) we get,

b2 s2 + b1 s + b0 = c1 (s + λ2 )(s + λ3 ) + c2 (s + λ1 )(s + λ3 ) + c3 (s + λ1 )(s + λ2 ) (2.28)

To find the values of c1 , c2 & c3 from (2.28), replace S with different values.

Substitute s = −λ2 in equation (2.28) we get,

58
 Chapter 2. Modeling in Time Domain

b2 λ22 − b1 λ2 + b0 = c2 (−λ2 + λ1 )(−λ2 + λ3 )

b0 − b1 λ2 + b2 λ22
c2 = (2.29)
(−λ2 + λ1 )(−λ2 + λ3 )
Substitute s = −λ3 in equation (2.28) we get,

b2 λ23 − b1 λ3 + b0 = c3 (−λ3 + λ1 )(−λ3 + λ2 )

b0 − b1 λ3 + b2 λ23
c3 = (2.30)
(−λ3 + λ1 )(−λ3 + λ2 )
Substitute s = 0 in equation (2.28) we get,

b0 = c1 (λ2 λ3 ) + c2 (λ1 λ3 ) + c3 (λ1 λ2 )

b0 − c2 (λ1 λ3 ) − c3 (λ1 λ2 )
c1 = (2.31)
(λ2 λ3 )
From equation (2.26) we can write output Y (s) as,

c1 c2 c3
Y (s) = U (s) + U (s) + U (s) (2.32)
(s + λ1 ) (s + λ2 ) (s + λ3 )
Equation (2.32) can be written as,

Y (s) = c1 X1 (s) + c2 X2 (s) + c3 X3 (s) (2.33)

From equation (2.33), by taking inverse laplace transform we get,

U (s)
X1 (s) =
(s + λ1 )
X1 (s)(s + λ1 ) = U (s)
ẋ1 (t) = −λ1 x1 + u(t) (2.34)

U (s)
X2 (s) =
(s + λ2 )
X2 (s)(s + λ2 ) = U (s)
ẋ2 (t) = −λ2 x2 + u(t) (2.35)

59
 Chapter 2. Modeling in Time Domain

U (s)
X3 (s) =
(s + λ3 )
X3 (s)(s + λ3 ) = U (s)
ẋ3 (t) = −λ3 x3 + u(t) (2.36)

Similarly taking inverse laplace transform of equation (2.33) we get

y(t) = c1 x1 (t) + c2 x2 (t) + c3 x3 (t) (2.37)

From equation (2.34), (2.35) & (2.36) we can write the state equation as,

State equation
      
ẋ −λ1 0 0 x1 1
 1      
      
 =  −λ2 0 x2  +  1  u
      
 ẋ2 0  
      
      
ẋ3 0 0 −λ3 x3 1

From equation (2.37) we can write output equation as,

Output equation
 
x
 1 
h i h i 
=
 
y c1 c2 c3  x 2 
 
 
x3

60
 Chapter 2. Modeling in Time Domain

2.9.2 Block diagram representation of diagonal canonical

form–Case 1

ẋ1 R x1
c1
-

λ1

u ẋ2 R x2 y
c2
-

λ2

ẋ3 R x3
c3
-

λ3

Figure 2.5: Block diagram representation of diagonal canonical form–Case 1

2.9.3 Diagonal canonical form – Case 2

Consider the nth order dynamical system,

dn y dn−1 y dy dm u dm−1 u du
an n
+ a n−1 n−1
+ . . . + a 1 + a 0 y = b m m
+ b m−1 m−1
+ . . . + b1 + b0 u
dt dt dt dt dt dt
Above equation can be written in s domain as,

Y (s) bm sm + bm−1 sm−1 + . . . + b1 s + b0

G(s) = = (2.38)
U (s) an sn + an−1 sn−1 + . . . + a1 s + a0
Case 2: Proper transfer function (m = n)

Y (s) b3 s 3 + b2 s 2 + b1 s + b 0
G(s) = = 3 (2.39)
U (s) s + a2 s 2 + a1 s + a0

61
 Chapter 2. Modeling in Time Domain

Equation (2.39) can be written as:

Y (s) b3 (s3 + a2 s2 + a1 s + a0 ) − (b3 a2 s2 + b3 a1 s + b3 a0 ) + b2 s2 + b1 s + b0

=
U (s) s3 + a2 s2 + a1 s + a0
(2.40)

s2 (b2 − b3 a2 ) + s(b1 − b3 a1 ) + (b0 − b3 a0 )

 
Y (s)
= b3 + (2.41)
U (s) s 3 + a2 s 2 + a1 s + a0
Equation (2.41) can be written as:

 
Y (s) c1 c2 c3
= b3 + + + (2.42)
U (s) (s + λ1 ) (s + λ2 ) (s + λ3 )

 
Y (s) c1 (s + λ2 )(s + λ3 ) + c2 (s + λ1 )(s + λ3 ) + c3 (s + λ1 )(s + λ2 )
= b3 +
U (s) (s + λ1 )(s + λ2 )(s + λ3 )
(2.43)
Comparing numerator term of equation (2.41) and (2.43) and substitute s = −λ2

we get,

(b0 − b3 a0 ) − λ2 (b1 − b3 a1 ) + λ22 (b2 − b3 a2 )

c2 = (2.44)
(−λ2 + λ1 )(−λ2 + λ3 )
Comparing numerator term of equation (2.41) and (2.43) and substitute s = −λ3

we get,

(b0 − b3 a0 ) − λ3 (b1 − b3 a1 ) + λ23 (b2 − b3 a2 )

c3 = (2.45)
(−λ3 + λ1 )(−λ3 + λ2 )
Comparing numerator term of equation (2.41) and (2.43) and substitute s = 0 we

get,

(b0 − b3 a0 ) − c2 (λ1 λ3 ) − c3 (λ1 λ2 )

c1 = (2.46)
λ2 λ3
From equation (2.42) we can write output Y (s) as,

c1 c2 c3
Y (s) = b3 U (s) + U (s) + U (s) + U (s) (2.47)
(s + λ1 ) (s + λ2 ) (s + λ3 )

62
 Chapter 2. Modeling in Time Domain

Equation (2.47) can be written as,

Y (s) = b3 U (s) + c1 X1 (s) + c2 X2 (s) + c3 X3 (s) (2.48)

From equation (2.48), by taking inverse laplace transform we get,

U (s)
X1 (s) =
(s + λ1 )
X1 (s)(s + λ1 ) = U (s)
ẋ1 (t) = −λ1 x1 + u(t) (2.49)

U (s)
X2 (s) =
(s + λ2 )
X2 (s)(s + λ2 ) = U (s)
ẋ2 (t) = −λ2 x2 + u(t) (2.50)

U (s)
X3 (s) =
(s + λ3 )
X3 (s)(s + λ3 ) = U (s)
ẋ3 (t) = −λ3 x3 + u(t) (2.51)

Similarly taking inverse laplace transform of equation (2.48) we get

y(t) = b3 u(t) + c1 x1 (t) + c2 x2 (t) + c3 x3 (t) (2.52)

From equation (2.49), (2.50) & (2.51) we can write the state equation as,

State equation
      
ẋ −λ1 0 0 x1 1
 1      
      
 =  −λ2 0   x2  +  1  u
      
 ẋ2 0
      
      
ẋ3 0 0 −λ3 x3 1

63
 Chapter 2. Modeling in Time Domain

From equation (2.52) we can write output equation as,

Output equation
 
x
 1 
h i h i 
= c 3  x 2  + b3 u
 
y c1 c2
 
 
x3

2.9.4 Block diagram representation of diagonal canonical

form–Case 2

ẋ1 R x1
c1
-

λ1

u ẋ2 R x2 y
c2
-

λ2

ẋ3 R x3
c3
-

λ3

b3

Figure 2.6: Block diagram representation of diagonal canonical form–Case 2

64
 Chapter 2. Modeling in Time Domain

2.9.5 Advantages of diagonal canonical form

ˆ The system matrix [A] is always in diagonal form.

ˆ In this approach, each state equation is of first order irrespective of order of

the transfer function.

ˆ These equation can be solved independently.

ˆ Moreover, the decoupling of the state equation is also possible.

ˆ Due to diagonal nature of the system matrix, it is useful in determination of

state transition matrix, controllability & observability and stabilizability &
detectability.

2.9.6 Disadvantages of diagonal canonical form

ˆ Similar to companion form, diagonal canonical form are not the physical
variable of the system.

ˆ Therefore, they are difficult for measurement and control purpose practically.

Examples

E.g.1: Find the state space model for the given transfer function
Y (s) s2 + 2s + 2
= 3
U (s) s + 12s2 + 47s + 60

E.g.2: A state model of a system is

      
ẋ1 −p 1 x k

  = 
  1  +  1  u
ẋ2 −q 0 x2 k2
 
h i h i x1
y = 1 0  
x2

Find the values of p, q, k1 and k2 if the system transfer function is

Y (s) 5s + 15
= 2
U (s) s + 6s + 15

65
 Chapter 2. Modeling in Time Domain

2.10 State space model to Transfer Function

2.10.1 Introduction
ˆ Suppose for a plant, if we have developed a model, which is based on state
space. But, the main purpose is to design a controller (PID type controller).

ˆ Most of the PID controller design techniques are based on transfer function
approach.

ˆ In that case, there is need to convert state space model into transfer function
form.

ˆ In control theory, like model order reduction, most of the conventional techniques
are based on transfer function form.

ˆ In that case also, there is need to convert state space into transfer function.

2.10.2 Derivation

Given the state and output equations

ẋ = Ax + Bu (2.53a)
y = Cx + Du (2.53b)

Take laplace transform of Eq. (2.53) assuming zero initial conditions

sX(s) = AX(s) + BU (s) (2.54a)

Y (s) = CX(s) + DU (s) (2.54b)

Solving for X(s) in Eq. (2.54a) we get,

(sI − A)X(s) = BU (s) (2.55)

or

X(s) = (sI − A)−1 BU (s) (2.56)

where I is the identity matrix Substituting Eq. (2.56) into Eq. (2.54b) yields

66
 Chapter 2. Modeling in Time Domain

Y (s) = C(sI − A)−1 BU (s) + DU (s) (2.57)

Y (s) = [C(sI − A)−1 B + D]U (s) (2.58)

Solving Eq. (2.58) we get,

Y (s)
= G(s) = C(sI − A)−1 B + D (2.59)
U (s)

Eq. (2.59) can be written as,

Y (s) Adj[sI − A]
= G(s) = C B+D (2.60)
U (s) det[sI − A]

2.10.3 Determination of Adj [sI-A]

1. The adj [sI-A] is defined as the transpose of cofactor of [sI-A]

Adj[sI − A] = (cof actor[sI − A])T

2. Cofactor of [sI-A] is related to principle of a matrix. Suppose the minor of

element (aij ) is denoted by |Mij |. The |Mij | denotes the ith and jth column
of a determinant [A]. The cofactor of the element (aij ) is given as minor
(−1)i+j |Mij |.

3. Determinant of [sI-A]

2.10.4 How to find inverse of third order system?

Consider the third order system as

 
 a11 a12 a13 
 
[A] = 
 a21 a22 a23


 
a31 a32 a33

67
 Chapter 2. Modeling in Time Domain

Finding the minors of matrix [A]

       
a22 a23  a a23  a a22 
     
 k
 11 =   k12 =  21  k13 =  21  

    

 a32 a33  a31 a33  a31 a32  

       
 
a12 a13  a a13  a a12 
       
[A] =  k21

=   k22 =  11  k23 =  11  

    

 a32 a33  a31 a33  a31 a32  

       
 
a12 a13  a a13  a a12 
       

 k31 =   k32 =  11  k33 =  11  

    
a22 a23  a21 a23  a21 a22 

where kij is minor of aij

Cofactor of matrix = (−1)i+j Mij , where, i is row and j is column.
   
 m11 m12 m13   k11 −k12 k13 
   
Cofactor of [A] =  m21 m22 m23   =  −k21 k22 −k23
 

   
m31 m32 m33 k31 −k32 k33

 
 m11 m21 m31 
 
Adjoint of [A] = transpose of (Cofactor of (A)) = 
 m12 m22 m32


 
m13 m23 m33

2.10.5 How to find determinant of [A]?

The determinant of [A] can be calculated as

 
 a11 a12 a13 
 
[A] =   a21 a22 a23


 
a31 a32 a33

|A| = a11 (a22 a33 − a32 a23 ) − a12 (a21 a33 − a31 a23 ) + a13 (a21 a32 − a31 a22 )

68
 Chapter 2. Modeling in Time Domain

Examples

E.g.1: Find the transfer function model for the system given below
      
ẋ 0 1 x 0
 1  =   1  +   u
ẋ2 −1 −2 x2 1
 
h i h i x1
y = 1 0  
x2

E.g.2: Given the system defined by equation below, find the transfer function,
Y (s)
G(s) = U (s)
, where
 U(s)  isthe
 input and Y(s)
 is theoutput.
 
 ẋ1   0 1 0   x1   10 
      
 ẋ  =  0 0 1  x  +  0 u
 2    2   
      
ẋ3 −1 −2 −3 x3 0
 
x1 
h i h i 
y = 1 0 0   x2 

 
x3

69
 Chapter 2. Modeling in Time Domain

2.11 Time-Domain Solution of LTI State Equations

2.11.1 Introduction
ˆ The response of linear, time-invariant models expressed in the standard state
equation form is given as:

ẋ = Ax + Bu (2.61a)
y = Cx + Du (2.61b)

ˆ The solution proceeds in two steps;

1. first the state-variable response x(t) is found by solving the set of

first-order state equations, Eq. (2.61a)
2. and then the state response is substituted into the algebraic output
equations, Eq. (2.61b) in order to compute y(t).

ˆ As in the classical solution method for ordinary differential equations with

constant coefficients, the total system state response x(t) is considered in two
parts:

1. a homogeneous solution xh (t) that describes the response to an arbitrary

set of initial conditions x(0),
2. and a particular solution xp (t) that satisfies the state equations for the
given input u(t).

2.12 Solution of Homogeneous State Equations

2.12.1 Solution of the scalar differential equation
Before we solve vector-matrix differential equations, let us review the solution of
the scalar differential equation

ẋ = ax (2.62)

In solving this equation, we may assume a solution x(t) of the form

x(t) = b0 + b1 t + b2 t2 + . . . + bk tk + . . . (2.63)

70
 Chapter 2. Modeling in Time Domain

By substituting this assumed solution into Eq. (2.62), we obtain

b1 + 2b2 t + 3b3 t2 + . . . + kbk tk−1 = a(b0 + b1 t + b2 t2 + . . . + bk tk + . . .) (2.64)

If the assumed solution is to be true solution, Eq. (2.64) must hold for any t.

Hence, equating the coefficients of the equal powers of t, we obtain

b1 = ab0
1 1
b2 = ab1 = a2 b0
2 2
1 1 3
b3 = ab2 = a b0
3 3×2
..
.
1 k
bk = a b0
k!
The value of b0 is determined by substituting t = 0 into Eq. (2.63), or

x(0) = b0 (2.65)

Hence, the solution of x(t) can be written as

 
1 22 1 k k
x(t) = 1 + at + a t + . . . + a t + . . . x(0)
2! k!
= eat x(0) (2.66)

2.12.2 Solution of the vector-matrix differential equation

We shall now solve the vector-matrix differential equation

ẋ = Ax (2.67)

where x = n vector and A = n × n constant matrix

By analogy with the scalar case, we assume that the solution is in the form of a
vector power series in t, or

x(t) = b0 + b1 t + b2 t2 + . . . + bk tk + . . . (2.68)

By substituting this assumed solution into Eq. (2.67), we obtain

71
 Chapter 2. Modeling in Time Domain

b1 + 2b2 t + 3b3 t2 + . . . + kbk tk−1 = A(b0 + b1 t + b2 t2 + . . . + bk tk + . . .) (2.69)

If the assumed solution is to be true solution, Eq. (2.69) must hold for any t.

Hence, equating the coefficients of the equal powers of t, we obtain

b1 = Ab0
1 1
b2 = Ab1 = A2 b0
2 2
1 1
b3 = Ab2 = A3 b0
3 3×2
..
.
1 k
bk = A b0
k!
The value of b0 is determined by substituting t = 0 into Eq. (2.68), or

x(0) = b0 (2.70)

Hence, the solution of x(t) can be written as

 
1 22 1 k k
x(t) = I + At + A t + . . . + A t + . . . x(0) (2.71)
2! k!

The expression in the parentheses on the right hand side of the last equation

Eq. (2.71) is an n × n matrix. Because of its similarity to the infinite power series
for a scalar exponential, we call it the matrix exponential and write

1 22 1
I + At + A t + . . . + Ak tk + . . . = eAt
2! k!
In terms of the matrix exponential, the solution of Eq. (2.67) can be written as

x(t) = eAt x(0) (2.72)

Since the matrix exponential is very important in the state space analysis of linear

systems, which can be written in generalized form as

Matrix Exponential:
∞
X Ak tk
eAt =
k=0
k!

72
 Chapter 2. Modeling in Time Domain

2.12.3 Laplace transform approach to solution–Scalar case

Let us consider the scalar case:

ẋ = ax (2.73)

Taking laplace transform Eq. (2.73), we obtain

sX(s) − x(0) = aX(s) (2.74)

Solving Eq. (2.74) for X(s) gives

x(0)
X(s) = = (s − a)−1 x(0)
s−a
Taking inverse laplace transform of above equation gives the solution

x(t) = eat x(0) (2.75)

2.12.4 Laplace transform approach to solution–Vector-matrix

case

Let us consider the vector-matrix case:

ẋ(t) = Ax(t) (2.76)

Taking laplace transform Eq. (2.76), we obtain

sX(s) − x(0) = AX(s) (2.77)

Solving Eq. (2.77) for X(s) gives

X(s) = (sI − A)−1 x(0)

Taking inverse laplace transform of above equation gives the solution

x(t) = L −1 [(sI − A)−1 ]x(0) (2.78)

Note that

73
 Chapter 2. Modeling in Time Domain

I A A2
(sI − A)−1 = + 2 + 3 + ...
s s s
Hence, the inverse Laplace transform of (sI − A)−1 gives

A2 t2 A3 t3
L −1 [(sI − A)−1 ] = I + At + + + . . . = eAt (2.79)
2! 3!
From Eq. (2.78) and Eq. (2.79), the solution of Eq. (2.76) is obtained as

x(t) = eAt x(0) (2.80)

The importance of Eq. (2.79) lies in the fact that it provides a convenient means

for finding the closed solution for the matrix exponential.

2.12.5 State–Transition Matrix

We have derived the solution of homogeneous equation as

x(t) = eAt x(0)

Above equation can be written as

x(t) = Φ(t)x(0) (2.81)

where Φ(t) = eAt is known as state transition matrix.

State transition matrix: It performs a transformation on x(0), taking x from

initial state, x(0), to the state x(t) at any time t.

2.12.6 Important points about State–Transition Matrix

ˆ It is defined as the solution of the linear homogeneous state equation.

ˆ It is the response due to the initial vector x(0).

ˆ It is dependent on the initial state vector and not the input.

ˆ It is called as zero-input response since the input is zero.

ˆ It is called as free response of the system since the response is excited by the
initial conditions only.

74
 Chapter 2. Modeling in Time Domain

2.12.7 Properties of State–Transition Matrix

1. Φ(0) = eA0 = I

2. ẋ(t) = Φ̇(t)x(0) at t = 0, Φ̇(t) = A

3. Φ(t) = eAt = (e−At )−1 = [Φ(−t)]−1 or Φ−1 (t) = Φ(−t)

4. Φ(t1 + t2 ) = eA(t1 +t2 ) = eAt1 eAt2 = Φ(t1 )Φ(t2 ) = Φ(t2 )Φ(t1 )

5. [Φ(t)]n = Φ(nt)

6. Φ(t2 − t1 )Φ(t1 − t0 ) = Φ(t2 − t0 ) = Φ(t1 − t0 )Φ(t2 − t1 )

2.13 Solution of Non-homogeneous State Equations

2.13.1 Solution of the scalar differential equation
Let us consider the state equation

ẋ(t) = ax(t) + bu(t) (2.82)

Eq. (2.82) can be written as

ẋ(t) − ax(t) = bu(t) (2.83)

Multiply Eq. (2.83) both sides by e−at we get,

e−at [ẋ(t) − ax(t)] = e−at bu(t) (2.84)

LHS of Eq. (2.84) can be written as

 
d −at −at dx(t) −at −at dx(t)
[e x(t)] = e + x(t)(−a)e ⇒e − ax(t)
dt dt dt

d −at
[e x(t)] = e−at bu(t) (2.85)
dt
Integrating both sides of Eq. (2.85) from 0 to t

75
 Chapter 2. Modeling in Time Domain

Z t Z t
d −at
[e x(t)] = e−aτ bu(τ )dτ (2.86)
0 dt 0

Z t
−at
e x(t) − x(0) = e−aτ bu(τ )dτ (2.87)
0

Multiplying Eq. (2.87) by eat and rearranging we get,

Z t
x(t) = at
e x(0) +e at
e−aτ bu(τ )dτ (2.88)
| {z } 0
initial condition | {z }
response to input

2.13.2 Solution of the vector-matrix differential equation

Let us consider the state equation

ẋ(t) = Ax(t) + Bu(t) (2.89)

Eq. (2.89) can be written as

ẋ(t) − Ax(t) = Bu(t) (2.90)

Multiply Eq. (2.90) both sides by e−At we get,

e−At [ẋ(t) − Ax(t)] = e−At Bu(t) (2.91)

LHS of Eq. (2.91) can be written as

 
d −At −At dx(t) −At −At dx(t)
[e x(t)] = e + x(t)(−A)e ⇒e − Ax(t)
dt dt dt

d −At
[e x(t)] = e−At Bu(t) (2.92)
dt
Integrating both sides of Eq. (2.92) from 0 to t

Z t Z t
d −At
[e x(t)] = e−Aτ Bu(τ )dτ (2.93)
0 dt 0

76
 Chapter 2. Modeling in Time Domain

Z t
−At
e x(t) − x(0) = e−Aτ Bu(τ )dτ (2.94)
0
Multiplying Eq. (2.94) by eAt and rearranging we get,
Z t
At
x(t) = e x(0) + eA(t−τ ) Bu(τ )dτ (2.95)
0

Z t
x(t) = Φ(t)x(0) + Φ(t − τ )Bu(τ )dτ (2.96)
| {z } 0
initial condition | {z }
response to input

2.13.3 Laplace transform approach to solution–Vector-matrix

case

Let us consider the vector-matrix case:

ẋ(t) = Ax(t) + Bu(t) (2.97)

Taking laplace transform Eq. (2.97), we obtain

sX(s) − x(0) = AX(s) + BU (s) (2.98)

Solving Eq. (2.98) for X(s) gives

X(s) = (sI − A)−1 x(0) + (sI − A)−1 BU (s)

Taking inverse laplace transform of above equation gives the solution

x(t) = L −1 [(sI − A)−1 ]x(0) + L −1 [(sI − A)−1 BU (s)] (2.99)

x(t) = Φ(t)x(0) + L −1 Φ(s)BU (s) (2.100)

| {z } | {z }
ZIR ZSR

Examples

E.g.1: Obtain the time response of the following system

      
ẋ 0 1 x 0
 1  =   1  +   u
ẋ2 −2 −3 x2 1

77
 Chapter 2. Modeling in Time Domain

where u(t) is the unit step function occuring at t = 0.

E.g.2: A state variable system

      
ẋ1 0 1 x1 1
  =   +  u(t)
ẋ2 0 −3 x2 0

where u(t) is the unit step function occurring at t = 0 and initial conditions as
   
x (0) −1
 1  =  
x2 (0) 3

Determine the state transition matrix and the state transition equation

78
 Chapter 2. Modeling in Time Domain

2.14 Introduction to controllability and observability

ˆ The concepts of controllability and observability were introduced by Kalman.

ˆ They play an important role in the design of control systems in state space.

ˆ In fact, the conditions of controllability and observability may govern the

existence of a complete solution to the control system design problem.

ˆ The solution to this problem may not exist if the system considered is not
controllable.

2.15 Controllability

ˆ Controllability is concerned with the control of state variables in a finite time

period, by some unconstrained control signal.

ˆ If any state variable is independent of control signal, then it is impossible to

control this state variable and therefore the system is uncontrollable.

System States System States

(initial state) (final state)

x(t0 ) x(tf )

Control input

u(t)

ˆ A system is said to be completely controllable if it is possible to transfer the

system from any initial state x(t0 ) to any final state x(tf ) in a finite time
interval, t0 ≤ t ≤ tf , by an unconstrained control vector u(t).

2.15.1 Controllability Matrix

Consider a nth order plant whose state equation is

79
 Chapter 2. Modeling in Time Domain

ẋ = Ax + Bu (2.101)

is completely controllable if and only if the rank of controllability matrix QC is

n.

QC = [B AB A2 B ... ... An−1 B] (2.102)

Controllability Test

If |QC | 6= 0, then the system is controllable.

If |QC | = 0, then the system is not controllable.

2.16 Observability

ˆ Observability is concerned with solving the problems of reconstructing unmeasurable

state variables from measurable variables in the minimum possible length of
time.

ˆ A system which is not completely observable implies that some of its state
variables are shielded from observation.

ˆ It is possible to estimate the state variables only if the system is completely

observable.

80
 Chapter 2. Modeling in Time Domain

2.16.1 Observability Matrix

The system is said to be completely observable if every state x(t0 ) can be determined
from the observation of y(t) over a finite time interval, t0 ≤ t ≤ tf . The system is,
therefore, completely observable if every transition of the state eventually affects
every element of the output vector.
Consider a nth order plant whose state equation is

ẋ = Ax + Bu
y = Cx + Du (2.103)

is completely observable if and only if the rank of observability matrix QO is n.

QO = [C T C T AT C T (A2 )T ... ... C T (An−1 )T ] (2.104)

Observability Test

If |QO | 6= 0, then the system is observable.

If |QO | = 0, then the system is not observable.

Examples

E.g.1: Examine the controllability and observability of the system described by,
        
ẋ 1 1 x 0 x
 1  =    1  +   u; y = 1 0  1 
h i

ẋ2 −2 −1 x2 1 x2

E.g.2: Examine the controllability and observability of the system described by,
        
ẋ −1 0 x 0 x
 1  =    1  +   u; y = 1 2  1 
h i

ẋ2 0 −2 x2 1 x2

E.g.3: The state variable equations of a system are: ẋ1 = −3x1 − x2 + u, ẋ2 =
2x1 , y = x1 + u
The system is
(A) Controllable but not observable. (B) Observable but not controllable.
(C) Neither controllable nor observable. (D) Controllable and observable.

81
 Chapter 2. Modeling in Time Domain

2.17 Introduction to Eigenvalues and Eigenvectors

2.17.1 Introduction
ˆ The eigenvalues and eigenvectors of the system matrix play a key role in
determining the response of the system.

ˆ It is important to note that only square matrices have eigenvalues and eigenvectors
associated with them.

ˆ The word “eigen” comes from German and means “own”, while it is the
Dutch word for “characteristic”, and so this module could also be called
“Characteristic values and characteristic vectors”.

ˆ Eigenvalues and Eigenvectors have a number of properties that make them

valuable tools in analysis, and they also have a number of valuable relationships
with the matrix from which they are derived.

ˆ The eigenvalues and eigenvectors of the system determine the relationship

between the individual system state variables (the members of the x vector),
the response of the system to inputs, and the stability of the system.

2.18 Eigenvectors
Definition 1:
Let us say,
Ax = y

where, A is (n x n) matrix, x is (n x 1), y is (n x 1) vector.

From above equation, we can say that (n x n) matrix operator A, operate on

(n x 1) vector x, we get new transform (n x 1) vector y.

Definition 2:
The eigenvector can also be defined as the vector “x ” such that the matrix operator
(A matrix) transforms it to a vector λx This vector have the same direction in state
space as the vector x.

82
 Chapter 2. Modeling in Time Domain

Definition 3:
The eigenvectors of a system matrix [A] are all vectors, xi 6= 0, which under the
transformation of matrix [A] become multiples of themselves.

Axi = λi xi

2.19 Eigenvalues
2.19.1 Definition
The eigenvalues of the matrix [A] are the values of λi that satisfy Axi = λi xi for
xi 6= 0.
Why the eigenvalue of a system matrix is determined as |λi I − A| = 0?

Axi = λi xi
(λi I − A)xi = 0
xi = (λi I − A)−1 0
adj(λi I − A)
xi = 0
det(λi I − A)
Since xi 6= 0, a nonzero solution exists if

det(λi I − A) = 0

2.19.2 Properties of Eigenvalues

ˆ If the coefficients of A are all real, then its eigen values are either real or in
complex-conjugate pairs.

ˆ If λ1 , λ2 , . . ., λn , are the eigenvalues of [A], then

n
X
tr(A) = λi
i=1

That is, the trace of A is the sum of all the eigenvalues of A.

ˆ If λi , i = 1, 2, . . . , n, is an eigenvalue of A, then it is an eigenvalue of [A]T .

83
 Chapter 2. Modeling in Time Domain

1
ˆ If A is nonsingular, with eigenvalues λi , i = 1, 2, . . . , n, then , i = 1, 2, . . . , n,
λi
are the eigenvalues of A−1 .

ˆ If the matrix A is real symmetric (row and column elements are same), then
the eigenvalues are always real (no complex conjugate values).

ˆ The product of the eigenvalues of a matrix equals the determinant of the

matrix.

ˆ A matrix is singular if and only if it has a zero eigenvalue.

ˆ Eigenvalue can be zero but eigenvector cannot be a zero vector.

Examples

E.g.1: Find the eigenvalues and eigenvectors of the matrix,

 
−3 1
A =  
1 −3

E.g.2: For the system described by the state equation

      
 ẋ1   0 1 0   x1   0 
      
 ẋ  =  0 0 1    x2  +  0  u
   
 2  
      
ẋ3 0.5 1 2 x3 1
If the control signal u, is given by u = [−0.5 − 3 − 5]x + v then the Eigenvalues of
the closed loop system will be
(A) 0, -1, -2 (B) 0, -1, -3
(C) -1, -1, -2 (D) 0, -1, -1

References
1. Katsuhiko Ogata, Modern Control Engineering, 5th edition, PHI, 2010, ISBN:
978-0136156734.

2. Norman S. Nise, Control Systems Engineering, Wiley India Edition, 2018,

ISBN: 978-8126571833.

84