zyxw

zyxwvutsrqponm
zyxwvutsrq
Book Reviews

zyxwvutsrqpo
zy
zyxwvutsrq
PROCEEDINGS OF THE IEEE, VOL. 69, NO. 10, OCTOBER 1981

zyxwvutsrqponmlk
zyxwvuts
The following reviews were selected from those recently published in
various IEEE TRANSACTIONS and GrouplSociety Magazines and News-
letters. They are reprinted here to make them conveniently avaihble to
the many readers who otherwise might not have ready access to them.
Errch review is followed by an iakntification of its original source.
to indicate how, in most cases, thesetechniquesmay be derived by
1375

extension of the control techniques already well established for delay-

free systems” It is important t o reaJize that the author hasonlya
special class of timedelay systems in mind. This book does not deal
-
with timedelay systems of the form k(t) = u ( t ) + p ( t h) + yu(t) or
generalizationthereof, for which the stated objective would notbe
attainable. Rather, the timedelay systems treated consist of a delay-
free subplant in series with a constant delay. Depending on therelative
positions of the delay and the subplant, this structure covers both the
case of control delay and the case of measurement delay. Apart from
Stpbility md Robustness of M w b k Feedbock Sydm-M. G. Chapter 9, the book deals with single input, single output systems with
Sanfanov (Cambridge, MA: M.J.T. Press, 1980, 171 pp.). Reviewed by the plantto be controlled represented by G(s)e-m where G(s) is rational.
R R. Mohler, Oregon State University, Corvat%s, OR. This makes it possible to achieve the stated purpose.
The style of presentation is in accordance with the following quote
Though this text apparently is based on Safanov’s doctoral disserta-
tion, it seems broad enough in scope and sufficiently readable to merit
.
from the preface: “. .it has not been thought appropriateto offer what
might be called a rigorous treatment. It is helpful in engineering and in
consideration by control theoreticans, and advanced graduate students. the physical sciences to adopt a happy combination of physical insight
It is not likely that control system designers will find the book useful. and mathematical technique.” As a result, the text contains no t h e e
But, on the other hand, it could provide a base for more meaningful rems and proofs. Rather, the presentation is descriptive and tutorial
interpretations andpractical developments. with a fair sprinkling of examples illustrating the main techniques for
The book is timely in that effective modem design practice requires analysis and design. Each chapter is divided into several short sections.
more than another abstract reformulation of linear system theory. In- This has the advantage of easy referencing, but tends to break up the
deed, the author makes a worthy attempt in this direction. Unfortu- continuity of presentation. This was particularly felt in the extensive
nately, the problems are diffEult. As generalizations are attempted, the discussion of the Smith’s method in Chapter 5 and subsequent chapters
results sometimes become almost as difficult to apply as the basic defi- The prerequisites for reading the textare stated generally as “a famil-
nition itself. iarity of the control of delay-free systems.” Some knowledge of block
The author does an excellent job of making the text almost self- diagram manipulations,elementarycomplex analysis, and Laplace
contained with only a minimum background assumed in system theory, transforms is clearly necessary. An acquaintance with classical, single
topology, and functional analysis. This does not mean, however, that variable, frequency design methods (Bode, Nyquist, and root-locus) is
the uninitiatedreader will not find the text challenging or tedious. also essential. While these methods are reviewed for the delay-free case
Briefly, the monograph consists of four chapters. First, a short over- in Chapter 2, the treatment afforded is too brief to serve as a fust-time
view of the text andrelated research is presented on stability, robustness, introduction.
multiloop feedback stability and topological separation,LQG estimation
and control, and sensitivity. Then stability and robustness are analyzed Theauthor also provides an introduction to optimal control, ob-
in terms of a new form of topological separation of system function servers, filers, sensitivity, adaptivecontrol,and multivariable control
spaces which have a frequencydomaininterpretation, leading to a theory, claiming “that the central problem of the controlof timedelay
multivariable generalization of the circle criterion. It is shown that a systems is a useful peg on which such an introduction can be hung.”
Lyapunov function for a stable system naturally induces the conicity The coverage and discussion of these topics are, of necessity, very brief,
conditionsforthe function-spaceseparation. Similarly, themethod and it would have been helpful to readers unfamiliar with these topics
suffers some of theSame drawbacks as Lyapunov’s direct method. if more tutorial references had been cited for the delay-free case.
In Chapters 3 and 4, LQG stability and robustness are analyzed for The book consists of 11 chapters, a list of 112 cited references, an-
the continuous and the discrete cases respectively. I n particular, it is other list of 102 references not cited in the text, and a set of 11 exam
shown that a suboptimal constant-gain extended Kalman-Bucy filter is problems with solutions and/or hints used by the author at theUniver-
sity of Bath. Chapter 1 discusses how delays occur in measurement and
robust against controller design approximations with respect to stability in control in physical systems. Chapter 2 reviews Laplace transformation
(but not necessarily performance). A separation principle for a special
and z-transforms for continuous, discrete, and sampled (mixed) systems,
class of non-linear control systems with additive white state noise and
output noise is established with respect to stability (but notoptimality). andbrieflyoutlines the Bode diagram, Nyquist plot, and root locus
method for delay-free systems. The outlines are indeed very brief, e.g.,
The final two chapters are probably the most interesting. But no dis-
Bode diwams are given only 16 lines of text with no illustration.
cussion is given to optimal nonlinear fdtering and control, which has Chapter 3 analyzes timedelay systems of the formKC(s)e-m with unity
become apopular area ofrecent research. Examples which compare feedbackandintroduces the modifiedz-transform totreat systems
performances, (suboptimal vs. optimal), stability margins, and robust- with sampling in some detail. Examples of the extensions of Bode and
ness for nonlinear cases as well as extended linear perturbation cases Nyquist methods to delay systems are given and approximation tech-
would be most helpfuL niques for e-m, including Pad6 approximants, are discussed. Chapter 4
In conclusion, I am pleased to recommendthismonograph t o the considers stability, mostly by determining conditions for sustained os-
appropriate controltheoreticans. cillations for low-order choices ofsubplant GO). Chapter 5 explains
the Smith’s principle and the design method by the same name. The
sensitivity concept, introduced by the sensitivity-point method in the
Reprinted fromIEEE Circuits and Systems Magazine, March 1981.
next chapter, is then applied to the Smith’s method. The model-refer-
ence idea inherent in the Smith’s method and similar designs motivate
Control of Time-Delay Systems-John E. Marshall (IEE Control Engi- the discussion of adaptive control in Chapter 7. Chapter 8 features a
neering Series No. 10. Stevenage, U.K.: Peter Pereghus Ltd., and Pis- short overview of optimal control for Smith’s type design and Chapter
cataway,NJ: INSPEC Dept, IEEE Service Center, 1979, 237pp., 9 discusses multivariable control, though mostly for the delay-free case.
$26.00). Reviewed by Arild Thowsen, Zow State Universiv, Ames, Chapter 10 provides brief mention of Luenberger observers, K h a n f&
IA. ters, nonlineardelay systems, and self-tuning controllers,and derives

zyxwvutsrq
The Institution of Electrical Engineers published the first volume of
itsControl Engineering Series in 1976. The present bookby J. E.
Marshall is the tenth andlatest volume in this continuing seriesintended
for postgraduate students and professional control engineers.
The purpose of this introductory text on control of timedelay sys-
tems is, according to the author, “to present a systematic account of
the control methods available for timedelay systems, and in particular,
the minimum variance controller and a linear quadratic controller for
discretetime system with delay. Chapter 11, filling more than onesixth
of the text, discusses the relative merits of digital, hybrid, and analog
computationfor simulationof timedelay systems, and gives many
examples (in considerable detail) ofdigital and hybrid computation
using methods from thepreceding chapters.
Overall, the book provides a good overview of past work on the con-
trol of delay systems modeled by G(s)e--. Due to the extensive scope