This introduction to 'Industrial Digital Control Systems' serves two purposes. Firstly it gives a history of computers in Control and discusses the advantages of using them. Secondly it gives an overview of the digital control systems theory to be presented in the later chapters.

This chapter is concerned with the realisation of simple controller by digital means. Controllers may be designed ab initio in the digital domain and this approach has much to commend it. This technique works very well for systems where the sampling rate ca be made fast compared to the basic system time constant and in particular for process control systems which are usually very slow. Process control is characterised by systems which are relatively slow and complex and which in many cases include an element of pure time delay.

The application of state-space concepts to control engineering developed in the decade from 1950 and gave rise to “modern control theory”. Previous formulations of control theory relied on concepts related to transfer functions; the output/input relationship of a system expressed as a ratio of Laplace transforms of the input and output signals. Allied to this the powerful concept of frequency response, which had its roots in communication theory, formed the foundation of “classical control theory”. In this the notion of a “signal” was basic and the control system was regarded as a signal processing element. Modern control theory took a more detailed view of the internal structure of a system, and the many variables associated with it, so that the concept of “state” replaced that of the “signal” as the primary interest. This change of view was necessary to come to terms with problems of time optimal control and non-linear system stability. A consequence of this is a shift of view from the transfer function as an operator on signals to the matrix as an operator on states. The system response is seen as a progression of changes of state evolving through time.

Self-tuning control has been concerned with the sub-optimal (asymptotically optimal) control of noisy linear time-invariant systems of known order and delay but unknown parameters. A self-tuning regulator consists of three components: an identifier, a control synthesis algorithm and a feedback compensator.

As the need for industrial efficiency and competitiveness increases, so there is a growing interest in optimization and optimal control as a mechanism to aid in their achievement. There are, of course, related increases in the interest and need for mathematical modelling tools and algorithms to underpin these developments but these ideas are not within the brief of this chapter. It is, however, worth underlining the increased need for the mathematical technologies and investment in training in these areas.

This chapter presents a review of particular techniques that may be employed for the modelling and analysis of industrial processes and for the design of control systems to regulate those processes. These techniques have been integrated into a package of computer hardware and software that is now being marketed by Vuman Ltd. VUMAN stands for the “Victoria University of Manchester” and the company is wholly owned by Manchester University. The package is termed the “Plant Analysis System” or PAS for short. It has been successfully applied to a number of processes in a variety of industries.

In many application areas, processing requirements for digital control systems, such as execution time and algorithm complexity, have increased dramatically. For a growing number of real-time control applications, then, conventional single-processor systems are unable to satisfy the new demands for increased speed, greater complexity and greater flexibility. This chapter explores alternative architectures, specifically parallel processing architectures, and, through case study examples, demonstrates some strengths and shortcomings of the INMOS transputer.

Artificial intelligence (AI) techniques are increasingly being used throughout the field of process control where the applications benefiting from these techniques range from measurement (sensor) validation and control system design, to control loop supervision and tuning. This chapter concentrates on the on-line application of knowledge engineering techniques in real-time to achieve an improvement in process performance. Current trends in the application of expert system in industry have been outlined in a recent survey carried out by Sangiovanni and Romans (1). One of their more significant findings was that more than 30% of Expert Systems are used for the diagnosis of process or instrumentation faults whereas less than 5% of applications are concerned with process control. This is not surprising as real-time knowledge-base software has only recently become available and even so is still in the early stages of its evolution. This chapter attempts to overview the advantages of knowledge based systems (KBS) in order to satisfy the conflicting goals of maintaining process stability whilst meeting the demands of process operation. The difficulty in achieving 'good' control system performance results from having to accommodate inaccuracies (uncertainty) and changes (variations) between our theoretical understanding of the process and the actual process itself. The goal of current research is to improve the performance and operability of closed-loop systems where the structure of the solution is to be as general as possible and which naturally takes the form of a real-time KBS combining the knowledge of both control engineers and process engineers.

This case study describes the implementation of a number of discrete algorithms phase advance, PID, velocity feedback, state feedback using observers and finite-time settling controller to control an electro-mechanical system.

The work described in this chapter is part of a continuing study on the application of optimal control and filtering theory to CLOS guidance of a BTT missile. Direct application of optimal control theory is difficult because of the nonlinear nature of the system dynamics (Roddy and Irwin (8, 9)). The use of two identical, single-plane compensators for vertical and horizontal motion is a practical alternative and does facilitate linear design. However the three-dimensional CLOS guidance system is suboptimal in several respects; the roll response, although fast, is not instantaneous and the commanded pitch and roll motions combine to induce some yaw motion. A more sophisticated autopilot (McConnell and Irwin (10)) would attempt to suppress such cross-coupling. Furthermore, although the quadratic cost J penalises large acceleration demands, it does not guarantee that demanded elevator angles will not lie outside the physical range of the missile. Some of these problems can be alleviated by a careful choice of roll-loop compensator (Roddy et al (11)).

This chapter presents two case studies to illustrate the VUMAN approach to control system design and to demonstrate the practicality of the mathematical procedures that are reviewed in Chapter 11. All of the design results presented have been established using the VUMAN 'Plant Analysis System' package that is described in some detail in Chapter 11. It is necessary for the reader to have studied Chapter 11 in some detail in order to properly appreciate this chapter, since frequent reference is made throughout to aspects of Chapter 11.