The Application of On-Line Expert Systems to Supervisory Process Control

Abstract

The use of computers in the process industry is set to increase. However, the implementation of conventional computer technology is not adequate for executing the complex supervisory tasks. This paper speculates upon the use of on-line expert systems which have been developed for working in non-linear, non-algorithmic problem domains. Expert systems for supervisory control are explored, and the potential benefits for optimisation and improved accuracy of control are outlined. The complexity and multi-tasking nature of real-time supervisory control points to an expert system structure which is distributed. A modular expert system design is proposed and the potential benefits for the approach are evaluated.