A Rule Based Inference Method for Supervision of Self-Tuning Controllers by Using Microcomputers

Abstract

Practical designs and applications of self-tuning controllers have shown that besides the parameter estimation and control algorithms a significant amount of heuristic logic must be considered. Moreover, self-tuning controllers must also beable to handle violations of pre-conditions, and numerical problems and bursts observed in recursive parameter estimation algorithms. In fact, in all real processimplementations, some supervisory functions must be considered to avoid unacceptable closed loop behaviour. This paper describes the structure of an expert supervision level which can perform intelligent functions. Simulation results with two supervision functions have shown that a good overall behaviour is obtained.