You are on page 1of 2

Book Reviews 849

Automatic Control Systems (Fifth Edition)*


Benjamin C. Kuo

Reviewer: PROFESSOR T O R K E L G L A D function. The interpretation of Bode, Nyquist and Nichols


Department of Electrical Engineering, Link6ping University diagrams is dealt with in great detail.
of Technology, S-581 83 Link6ping, Sweden. The design methods described include classical techniques
based on the root locus, and compensation based on the
Bode and Nichols diagrams. They are supported by
THE BOOK UNDER REVIEW is a classic introductory text in
Automatic Control, this being the fifth edition. As the numerous examples. The state space methods are less well
represented. Pole placement via state feedback is treated but
subject has grown and matured, the task of writing such a
book has in a sense grown more difficult. This is mainly not observers. The use of integrators with state feedback is
because the number of topics that are considered essential discussed, however. This is a topic that is often missing in
have grown, so that much material has to be covered even in introductory texts on state space methods. From the
engineering point of view it is essential of course, since so
a basic textbook. The size of the book, 720 pp. bears
testimony to this. Even so, many things have to be excluded many control systems need the integrator to work at all.
and one major task of the book writer is to decide what More advanced topics such as linear quadratic control are
should be included and what should be left to more advanced excluded.
courses/books. The treatment of state space has a problem that is
Essentially there are three things you need for a systematic probably well known to many people who have taught
design of a controller. You need a model of the system, automatic control courses. A lot of time and effort is spent
methods for analyzing the model and methods for developing the machinery of state space theory, including
synthesizing a controller for it. To show how a model is matrix exponentials, controllability and observability, etc.
obtained, the author has included a chapter on modeling of This machinery is not used to such an extent that it seems
physical systems. The modeling is done in state space and the justified to the student however. The only design tool is pole
concept of state is motivated from a physical point of view. placement using state space techniques. The topics that really
Several typical servo applications illustrate the modeling show up the possibilities, like optimal control, Kalman
process. The other possible way of obtaining models, via filtering, time variable systems, nonlinear systems, are not
identification, is not mentioned in the book. touched upon. To include all the topics in an elementary
When it comes to the analysis of models, there are many textbook is obviously impossible. The solution of excluding
state space theory would completely leave students badly
questions. Should the analysis be done in continuous time or
discrete time, in the time or in the frequency domain, using equipped for deeper studies in control. Perhaps the problem
state space or input-output models? What makes control is insolvable. Maybe the inclusion of state reconstruction
systems special from the continuous-discrete time point of with observers, in Kuo's book, would have rounded off the
view is that they are usually hybrids. A physical plant that state space presentation nicely.
evolves in continuous time is controlled by a computer that The strength of the book probably lies in its detailed and
operates in discrete time. Thus it is difficult to give even an well thought out explanations of central topics like the root
introductory control course without touching on both cases. locus and the Nyquist criterion. These are accompanied by
In this book the author has concentrated on the continuous many figures showing the relationships being discussed. The
time description that is natural from the point of view of explanations of some of the state space ideas seem less
process modeling. In Chapter 5 there is a description of the complete. In the discussion of the criterion of observability
sampling of a state space model for piecewise constant for instance, the observability test matrix crops up without
inputs. The description of the Nyquist criterion in Chapter 9 any explanation. The author does not give a mathematical
also contains a discussion of its discrete time counterpart. proof of the result, which is reasonable, considering the aim
The preference for frequency or time domain analysis has of the book (a reference is given where a proof can be
shifted back and f o r ~ during the development of Automatic found). However, it would have been possible to gi~e some
Control. After the period of classical frequency analysis with motivation, for instance by showing how the state can be
Bode, Nyquist and Nichols there came the state space era computed from the output and its derivatives.
which was followed by a renewed interest in frequency Throughout the book there is a conscious effort of tying
methods, spurred in part by robustness problems. One together the theory with physical reality, mainly with
consequence of this is that the control engineer, who wants examples and the large collection of problems. In this
to deepen his knowledge by further reading, has to know context it is of interest for the student to know when he can
both aspects. As a consequence the present book treats both expect the models to agree with reality and when he cannot.
time and frequency domain analysis in detail. From the One major cause of such a disagreement is the presence of
discussion of modeling in the state space it comes natural to nonlinearities. These do not show up in the linear model, of
develop time domain analysis using the state space course. It would lead too far if a detailed discussion of
formalism. This includes the state transition matrix, nonlinear system theory were included in a book like this. At
controllability and observability, canonical forms and related the same time one feels that the reader should be made to
issues.
realize that the world is not linear. Kuo has solved the
There is a detailed discussion on the relations between the problem by giving a discussion of nonlinear models and
time response and the poles of the system. This is followed linearization around an operating point. Perhaps it would be
up with a discussion of the root locus technique. The nice to have some discussion of nonlinear effects in the
frequency domain analysis includes a discussion of the design chapters. Reading, for example, about pole
connection between system properties and the transfer placement, and looking just at the mathematics, one might
get the idea that any system can be made as fast as desired,
while in practice there is a limitation due to saturation of the
* Automatic Control Systems (Fifth Edition), by B. C. signals, even if there is no other nonlinear effect.
Kuo. Prentice-Hall, Englewood Cliffs, New Jersey (1987). Taking an overview one feels that on the whole the book
736 pp. US $24.95. £15.95. represents a good choice on what should and should not be

AUTO 24:6-H
850 Book Reviews

included in an introductory text. Certainly the core material physics in 1970 and the Ph.D. degree in automatic control in
is well covered and accompanied by the examples and 1976, both from the Lurid Institute of Technology, Lund,
problems that are needed for understanding. Obviously Sweden. Since 1977 he has been employed as a Senior
Kuo's book must be considered when one chooses the text Lecturer in Control at the Department of Electrical
book for a course in Automatic Control. Engineering, Link6ping University in LinkOping, Sweden.
His research interests include nonlinear systems, optimal
About the reviewer control and mathematical programming.
TORKEL GLAD was born in Lund, Sweden in 1947. He
received the M.S. degree (Civilingenj6r) in engineering

You might also like