Real-time computer control: An introduction PDF

Preview

Summary

Automatica, Vol.27, No. 2, pp. 427-434, 1991 PergamonPressplc. Printedin GreatBritain. InternationalFederationof AutomaticControl B o o k Reviews Real-time Computer Control: An Introduction* Stuart Bennett Reviewer: L. MOTUS derivative calculations, z-transform. Example 4.5 contains an Institute of ...

Description

Automatica, Vol.27, No. 2, pp. 427-434, 1991

PergamonPressplc. Printedin GreatBritain. InternationalFederationof AutomaticControl

B o o k Reviews

Real-time Computer Control: An Introduction* Stuart Bennett derivative calculations, z-transform. Example 4.5 contains an error (nextSamplelnterval pro nextSampleTime). Chapter 5, Design of Real-time System, gives a reasonable survey of the hot topics of software engineering environments for computer control systems. I cannot fully agree with separating hardware and software designs before requirements analysis; perhaps it would be better to prove consistency and noncontradiction of the system definition (requirements) first. In many cases it would be difficult to fix a feasible hardware configuration before we have complete requirements on the system behaviour. Also, in control systems it may often happen that it is necessary to implement some part of software by using dedicated hardware and again, this kind of a decision depends on the control algorithms and on the plant dynamics. I doubt that the system definition (see Fig. 5.1) and the indicated interactions (see Fig. 5.1 and Fig. 5.2) are sufficient for that. At least these problems should have been mentioned in this chapter. Another problem worth mentioning in this chapter is connected with the multiple use of the system/software/ hardware abstract model. Traditionally the abstract model is used for providing a better survey of the system/software/ hardware to the designer in a hope that he/she can find discrepancies or errors in the description. Today the use of formal methods for discovering errors gains more and more popularity. Application of formal methods to system descriptions, in their turn, imposes new requirements to abstract models. This problem is not discussed in the book. Chapter 6, Operating Systems, gives a good explanation of the basic parts of these systems. This explanation is necessary and sufficient for understanding the functioning of a real-time system. A large part of the information obtained from the design description (see the previous chapter) is often used to determine or modify some of the operating system properties (e.g. scheduling, queue lengths, etc). The summary of this chapter just mentions the problem. I think students would need a more detailed discussion of this subject. Chapter 7, Concurrent programming, gives a pragmatic overview of parallel execution of programs--described are methods and primitives necessary for implementing multitasking (mutual exclusion and intertask communication problems). Presentation of methods and primitives is clear and understandable and is supplied with a number of examples. The given amount of information is sufficient for building not too demanding control systems. However, some of the practising engineers may have encountered more sophisticated concurrent programming problems, e.g. timing problems in intertask communication, or really concurrent implementation of some of the algorithms. For those it would be useful to know that concurrent programming means much more than is presented in this chapter. The heading of Chapter 8, Real-time Languages, is quite a confusing one. Not much is said about real-time languages in it, in fact, the next chapter handles them more thoroughly. This chapter is about requirements to languages for programming large systems. Chapter 9, Programming Languages, surveys some of the languages used for implementing computer control systems (Basic, Fortran, Pascal, Coral 66, RTL/2, Modula-2, Ada). I would have added PEARL to the list of surveyed languages. PEARL is used on the European continent and has a nice design supporting system (EPOS)--it provides a good example of systematic approach to the development of real-time systems. On the whole the book presents a well-balanced collection of topics important for designing and implementing computer

Reviewer: L. MOTUS

Institute of Cybernetics, Computer Division, Akadeemia tee 21, Tallinn 200108, Estonia. THIS BOOK is much better than I expected it to b e - - I was sure that it was another book on control algorithms, emphasizing now and then the necessity of very rapid computations. Fortunately enough, this was not the case. I think that the book presents a nice collection of knowledge areas, necessary to specify, design, implement, debug and test a real-time computer control system. The depth of presentation is suitable for the audience intended for the book--the final year undergraduate students and practising engineers. Still, I tend to think that an engineer can learn more from this book than a student: implicit connections between different knowledge areas may be a little bit difficult to discover without a practical control system background. In the following, the book is reviewed chapter by chapter and subjective merits and demerits are listed for each chapter. Chapter 1, Introduction to Real-time Systems, is perfect, especially the interpretation of real-time--in most cases real-time means that two or more systems (one being a computer-based system) with different inner time concepts are forced to cooperate by providing a common understanding of time. It was so refreshing to see that the widespread misconception of real-time as being almost equivalent to super-performance is no longer taught to students. Chapter 2, Concepts of Computer Control, seems to be not quite balanced. The attention has been biased towards explaining different control strategies (sequence control, direct digital control, supervisory control), whereas the problems of control system as a whole (centralized computer control, hierarchical control system, distributed control system) are discussed superficially. In addition, I am not quite happy with the notion of distributed system--in the book it is interpreted more like distributed computing. I would personally use the term "distributed control system" as a common name for a system that may contain both hierarchical control and distributed computing elements. This chapter would have been a suitable place for explicitly stating how the following chapters are related to each other in the process of building a control system. However, there is no such statement in the book. Chapter 3, Computer Hardware Requirements for Real-time Applications, is a very good explanation of how computer hardware functions (including also process interface and data communication in distributed system). The content of this chapter is closer to the description of the state-of-the-art of computer hardware than to the requirements. I would have expected more thorough discussion of standards; 24 lines is certainly not sufficient. Chapter 4, DDC Control Algorithms and their Implementation, is another excellent part of this book. The reader is systematically introduced to the differences of coding a data processing algorithm and programming a real-time system. Slightly out of line of this book seem to be the details of DDC algorithms, e.g. improved forms for integral and

* Real-time Computer Control: A n Introduction by Stuart Bennett, in Series in Systems and Control Engineering (Series editor: M. J. Grimble). Prentice Hall International, Hemel Hempstead, U.K. (1988). ISBN 0-13-762485-9, 362 pp, $82.95.

427

428

Book Reviews

control systems. It is readable and the presentation is clear. The basic drawback of the book is that it does not merge the different areas of human knowledge into one; this is left for the reader. I am afraid that this task is not so easy for an undergraduate student. In spite of this drawback the book is one of the best textbooks introducing real-time computer control that f have read.

A b o u t the reviewer

Leo Motus has Ph.D. equivalents in stochastic control and software engineering. He has been Chairman of the fFAC working group on Distributed Computer Control Systems since 1987, is a Vice-chairman of IFAC TC on Computers, and is a candidate for the post of TC Computer Chairman for the period 1990-1993.

Computer Control of Machines and Processes* John

G. Bollinger

Reviewer: C. SCHMID Ruhr-Universit~it Bochum, Lehrstuhl fiir Elektrische Steuerung und Regelung, Postfach 102148, D-4630 Bochum 1, Federal Republic of Germany.

THIS BOOK is an excellent one for undergraduate and graduate students of all engineering disciplines interested in computer control, as well as being a superb reference book for industrial people attending refresher courses on this subject. One of its striking features is that little prior knowledge is demanded from its reader/user. Its spectrum of topics, which are of special importance in computer control, is indeed exceedingly broad. The book has naturally evolved from courses presented at the University of Wisconsin (Madison) by the authors since the mid-1960s. In its more than 600 pages, the book covers all the relevant aspects of computer control. The twelve chapters essentially give a nonmathematical introduction. After the brief introductory Chapter 1, which sets the rest of the book in proper perspective by dealing with the history of computer control and the explosion in the application of electronic technology since the 1960s, it goes on to present chapters on elements of discrete-time modelling, system response generation, discrete-controller design, control computer hardware and software, computer interfacing, sensors, command generation, sequential logical control, process modelling, analysis and design. Finally, the book closes with an appendix on the state-variable approach. Chapter 2 starts with a discussion on basic process types and develops the idea of representing components of closed-loop systems using difference equations and discretetime transfer functions, without applying any transformation techniques (which will be first introduced towards the end of the book in Chapter 11). After reading this chapter, one is on the level to understand system input-output behaviour and to analyse the stability of a system. However, the discussion of stability is different from the standards of control theory! As the stability analysis is based on the roots of the characteristic equation in terms of the backward-shift operator, all discussions about stability are treated in a domain mirrored at the unit circle. This may cause the novice reader to be confused when reading other books recommended in the bibliography attached to this chapter. Two fundamental concepts are introduced in Chapter 3: The sampling of signals generating time series, and the use of the time-shift operator to generate and determine system responses. The first design of a simple controller can be found in Chapter 4, which focuses on the desired closed-loop rezponses to specific inputs. The discussion about the selection of the sample period and the design of feedforward, cascade and noninteracting control in interactive plants * Computer Control o f Machines and Processes by John G. Bollinger and Neil A. Duffei. Addison-Wesley, Reading, MA (1988). fSBN 0-201-10645-0, $53.75.

and Neil A.

Duffei

completes the first part of this book. These four chapters give an excellent introduction to discrete-time control. The second part of this book deals with the computer. Chapter 5 is dedicated to computer hardware and software. It presents important aspects of computer architecture and operation in a way that is independent of computer manufacturer or model. This includes binary logic, basic computer hardware, the concepts of instructions and data, input/output, interrupts and programming at the assembly language level. A set of high-level language procedures is defined such that the principle of closed-loop computer control can be shown at the end of the chapter. This chapter concludes with a discussion of how closed-loop control functions can be organized on a control computer. The material and examples are well elaborated and their sequence reflects the various levels of abstraction in which designers of computer-control systems must carry out their work. Linking to external devices is the main topic of Chapter 6. The range is from analogue conversion to address decoding, device selection and interrupt interfacing. The short discussion here draws on a simplified computer architecture and is sufficient for understanding. Chapter 7 describes a spectrum of sensors that are often found in computer control systems for machines and processes. Chapter 8 adds some nonessential aspects of command signal generation in control. The implementation of logic control and the solution of Boolean equations are discussed in Chapter 9. Ladder diagrams, which are the means to describe the solution of logic on a computer, are treated. In addition, a number of design methods for logic control are described--including the use of flowcharts, switching tables and state diagrams. This chapter gives only a rough sketch about the principles of sequential control and the use of programmable logic controllers. Chapters 1-9 can be covered in one semester at an introductory level. The third part of the book portrays more control techniques, the crucial point being the application of more sophisticated system techniques to computer control. Chapter 10 reviews a number of approaches to process modelling--from physical modelling to a mathematical one. Step-response and least-squares techniques of process model identification are used. Through Chapter 11, where he can find the bases of transformation into the frequency domain, the reader will be able to gather much more insight into dynamics. The concepts of ideal samplers, hold elements and the representation of a sample sequence are also introduced here. After the analysis section in Chapter 12, the design of controllers in the frequency domain is discussed both for continuous- and discrete-time systems, illustrating their similarities and differences. An appendix gives a brief review of state-variable methods for the analysis and design, including state estimators for continuous- and discrete-time systems. Throughout the textbook, technology-based information has been strictly avoided. The simplified computer hardware which is used allows fundamental concepts to be illustrated,...