SOFSEM 2001 28th Annual Conference on Current Trends in Theory and Practice of Informatics November 24 - December 1, 2001 Piestany, Slovak Republic, Europe

Knowledge-based Control Systems

by Simon Lambert

The monitoring and control of industrial devices, systems and installations has attracted attention from the field of knowledge-based systems for a number of reasons. The problem is intrinsically knowledge-intensive, needing awareness of the behaviour and interactions of physical components; there are complexities such as time constraints, partial and qualitative information, and possibly a need for adaptation over time; and in many cases there is a need for a degree of human understandability in performance.

Many techniques have been applied to aspects of monitoring and control, including rule-based, model-based and case-based reasoning. At a higher level, architectural issues also arise, including the use of agents, customisability of tools across applications or sites, and remote diagnosis over the Internet. There are general issues of combining the knowledge-based approaches with numerical ones, and, in the opposite direction, cooperating effectively with human operators.

This talk will survey the background, problems and approaches to knowledge-based control systems, and will describe work done in a variety of applications, and particularly work in progress on the development of an adaptive and customisable control system for anaerobic waste treatment plants.