A model-object based expert system for real-time intelligent control of chemical processes

Abstract

A supervisory control expert system with object-based knowledge representation has been developed such that the general knowledge confined to classes and subclasses forms a core knowledge base which can be used as primitives for rapid prototyping of intelligent control systems for various chemical processes. The expert system is embedded in an application program and communicates with the control system structured in the form of control blocks. The expert system prototype is developed for retuning model-based controllers to improve the controlled behavior of a packed-bed chemical reactor system. The knowledge base of the model-object-based expert control system captures the complex structure and interdependencies of process control systems into a compact form that includes structure, declarative and heuristic relationships, and procedural and model based knowledge.<>