Aukmtolica, Vol. 32, No. 6. pp. 945-949.

19%
Copyright 0 1996 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
m5-1098/% $15.00 + 0.00

Book Reviews

Feedback Control Systems*

Reviewer: HARRY L. TRENTELMAN meaning that they illustrate the theory but have no physical
Research Institute for Mathematics and Computing Science, meaning). In fact, at the beginning of the book, two entire
University of Groningen, chapters are devoted to the subject of deriving higher-order
P.O. Box 800, linear differential equations representing models for all kinds
9700 AV Groningen, The Netherlands. of physical systems. I was very pleased to see the models in
the book being actually derived, instead of being put forward
The third edition of the book Feedback Control Systems is a without further comment as so often occurs in textbooks on
revised version of the first (1986) and second (1990) editions. control. In later chapters, the models derived in these two
The book contains an introduction to control theory for introductory chapters are used to illustrate the theory.
linear systems. The first main part of the book is meant as a I shall now give a more detailed description of the contents
text for a first course on control systems for advanced of the book. In the first chapter, a general introduction is
engineering students. Later chapters could be used for a given to control systems, including a discussion of the notion
more advanced course. of open-loop and closed-loop control. Using a simple
The main part of the book is concerned with what is example, a discussion is given of the fundamental features of
usually called the classical approach to control. In this control system design. The chapter also introduces Laplace
approach, models for systems (mostly first-, second- or transforms and the concept of transfer function.
higher-order linear differential equations obtained from Chapter 2 consists of numerous examples of modelling
first-principles modelling) are immediately put in their mechanical, electrical thermal and fluid systems using
associated transfer-function form and represented by higher-order, linear, constant-coefficient differential equa-
equations of the form tions and their corresponding transfer functions. The
approach to these examples is very nice: at the beginning of
Y = g(r)u, each subject area, first some elementary ‘building blocks’ are
where g(s) is the transfer function of the system. In the discussed. For example, in the treatment of thermal systems,
classical approach (also often called the frequency-domain first the building blocks ‘thermal resistance’ and ‘thermal
approach), the analysis of the control system, and the design capacitance’ are discussed. Next, these elementary building
of controllers is based on the properties of the underlying blocks are used to obtain models in a systematic manner. For
transfer function(s). The chapters dealing with the classical example, in the discussion on thermal systems, the two
approach are all restricted to single-input/single-output aforementioned building blocks are used to model a process
systems. control system and a heating system.
In the second part of the book, an introduction is given to Chapter 3 contains several examples of transfer function
the state-space approach to linear control systems. Instead of models of simple, physical, real-world control systems, like
representing system models by transfer functions, this the ubiquitous water-level control system, a hydraulic
approach is centered around the paradigm of representing servo-mechanism with a mechanical feedback, and a
the higher-order linear differential equation describing the pneumatic pressure regulator. Also, a couple of examples
system model by a model in state-space form: from aerospace engineering (attitude control of satellites and
spacecraft) are considered. The chapter also discusses system
z(t) = Ax(t) + bu(t), (1) simulation using operational amplifiers, and the use of flow
diagrams.
y(t) = Cx(t) + du(t). (2) The fourth and fifth chapters deal with performance of
feedback systems and the design of feedback controllers to
Different from the classical approach, which from the outset improve the system’s performance. An elementary discussion
shifts from variable time t to variable frequency s, the is given of the motivation for using feedback control. It is
state-space approach sticks to variable time: the state-space shown that feedback controllers are able to reduce the
approach works within the time domain. Analysis and design sensitivity of the system’s performance to modelling
of control systems in this approach take place by studying the uncertainty and exogenous disturbances, to reduce the
associated matrices A, b, C and d. steady-state error in the system’s step response, and to
Smaller parts of the book are concerned with the analysis improve the system’s transient response. All this is illustrated
and design of sampled-data control systems (also in the by means of examples of first- and second-order systems.
frequency domain), multivariable control systems and Using the example of the lightly damped harmonic oscillator,
nonlinear control systems. a discussion is given of features of the system’s step response
The book is extremely well laid out. Every chapter starts (overshoot, peak time, rise time and settling time) and the
off with a clear introduction in which the issues to be role these features play in the system’s performance. Several
addressed are described. At the end of each chapter, there is control paradigms (PI, PD and PID) are discussed, and their
a section containing conclusions. The material in the book is limitations are explained.
illustrated by numerous pictures, plots and flow diagrams. In Chapter 6 deals with the graphical technique of root locus
addition, the system properties and controller designs are in the analysis and design of control systems. Chapters 7 and
illustrated by a large number of very nice physical examples. 8 give an extensive discussion of the role of the system’s
In my opinion, the author of the book has done a great job frequency response in analyzing control systems and
here: the examples are sufficiently simple to really illustrate designing controllers. The graphical technique of polar plot is
the theory, but are hardly ever academic (‘academic’ introduced and the Nyquist stability criterion is formulated.
Also, a nice treatment is given of the graphical technique of
*Feedback Control Systems, 3rd ed., by John van de the Bode plot. The relations between certain features of the
Vegte. Prentice-Hall, Englewood Cliffs, NJ (1994). ISBN Bode plot on the one hand and the performance criteria of
0-13-016379-l. gain and phase margin on the other are explained. In this

945
946 Book Reviews

way, the system’s performance is expressed in terms of introductory course-and even for a more advanced
properties of the system’s Bode plot. This is used as a basis one-the book is really excellent. The presentation is very
for controller design. clear, and the concepts and ideas are illustrated extremely
At this point in the book, the treatment of the classical well by working them out for several interesting examples.
approach to feedback control is finished. The subsequent However, undoubtedly because of its first appearing in 1986.
Chapters 9 to 10 deal with the analysis and design of the book hardly makes contact with more advanced ideas.
sampled-data control systems. all from a frequency-do-main concepts, techniques and results that have been developed
point of view. In chapters 11 and 12, the state-space well over the last 10 years. Here, I think of ideas on system
approach to linear control systems is introduced. Chapter 11 uncertainty in terms of H, error bounds. on the notion of
starts with introducing the standard (SISO) state-space robustly stabilizing a system, on quantifying the performance
model that will be used. Furthermore, it contains a discussion of a system in terms of the H, norm of certain closed-loop
of state trajectories. stability, controllability, observability transfer functions, and on the use of weighting filters in order
and stabilizability. Chapter 12 deals with designing to achieve good performance over certain frequency ranges.
controllers, in particular controllers that achieve pole In my opinion, it can very well be argued that some of these
assignment and controllers that solve the so-called optimal new ideas and concepts should not be omitted even from a
regulator problem. This chapter also contains a discussion of first course on control systems. simply because they are so
dynamic observers. Chapter 13 deals with multivariable fundamental.
systems in the frequency domain (Nyquist array technique In conclusion, I think this book is very nice. It is very well
and Gershgorin bands). In Chapter 14, an introduction is suited to learn the classical basics on control, even outside a
given to nonlinear control systems (Popov and circle classroom. The book greatly benefits from the fact that it has
criteria). Finally, the book contains two appendices. In the gone through several updating rounds. Formulations and
first, some basic facts on vectors and matrices are collected. explanations are very clear, and the worked out examples are
The second appendix contains a description of executable excellent. The book is clearly a product of great
programs that can be found on the floppy disc accompanying craftmanship.
the book. This disc contains programs to plot for a given
system its unit step response, its root locus. its polar plot and About the reviewer
its Bode plot. Harry L. Trentelman was born in Voorst. The Netherlands
As I already mentioned, the first edition of this book in 1956. He has been Associate Professor of Systems and
appeared in 1986. The core of the book is meant as an Control in the Department of Mathematics of the University
introductory treatment of the classical approach to control of Groningen. The Netherlands since 1991. He is Associate
systems. More advanced material (on sampled-data systems, Editor of the SIAM Journal on Control and Oatimization.
on the state-space approach, on multivariable systems and on His Research interests are representation and’ control of
nonlinear systems) is provided in later chapters. For an linear systems.

Modeling of Dynamic Systems*

Lennart Ljung and Torkel Glad

Reviewer: J. STECHA observed real plant. The last part of the book is dedicated to
Czech Technical University. the utilization of mathematical models for system simulation
Faculty of Electrical Engineering, and model validation. The parts of the book will now be
Department of Automatic Control. described in a little more detail.
Charles Square 13, 12135 Prague, The first part, entitled ‘Models’. briefly describes the
Czech Republic. different types of models of dynamic systems. Input-output
and state-space system descriptions, models of signals in the
This book deals with different approaches to mode1 building time and frequency domains. continuous- and discrete-time
for physical processes and systems. Model building of models. and discretization of continuous systems are also
processes is one of the first stages of engineering work. discussed briefly. Many examples are given here, and
Model building is an art in which the model builder utilizes particular emphasis is given to the connections between
all of his or her knowledge and experience to determine just different concepts. This part is not intended to cover in detail
one of many ways to solve a problem. This creative process the different forms and properties of mathematical systems.
has its own procedures, and this is the content of the book. ‘Physical Modelling’ is the second part of the book. This
The book is appropriate for all those for whom it was part describes how to construct a mathematical model from
written, namely undergraduate students of all engineering physical laws in electrical circuits, mechanical systems, flow
faculties and people who are interested in mode1 building in systems and thermal systems. It is shown that there is an
different areas of engineering, the natural sciences. and in analogy between different physical systems, thus allowing the
economics and the social sciences as well. development of a systematic modelling procedure for a broad
The book has 360 pages, and is divided into four parts. The class of systems. This is done using bond graphs. I am
division of the material is natural. First, it is necessary to convinced that the model builder must be a specialist in the
understand the different mathematical forms of models. appropriate area and that a good mathematical model of
There are two basic approaches to model building- reality can only be created through the cooperation of the
analytical and experimental. The analytical approach, specialist and the system engineer. However, both must share
described in Part 2. uses physical laws to describe the the same language, and their knowledge must partly overlap
behaviour of a dynamic system in mathematical form. The in order to cooperate effectively. This is the aim of the book.
experimental approach, described in Part 3, uses measured The third part, entitled ‘Identification’, shows how to use
data from the process to obtain a mathematical model of the data for model building. Many estimating procedures are
given: estimation of transient response in the time domain.
frequency analysis and signal spectra estimation. Great
*Modeling of Dynamic Systems, by Lennart Ljung and emphasis is given to system parameter estimation, but only
Torkel Glad. Prentice-Hall. Englewood Cliffs, NJ (1994). one-shot methods are presented. Problems from the field of
ISBN 0-13-597097-O. design of identification experiments are also mentioned. In