An Expert System Towards Multi-attribute Fuzzy Decision-making Of Controller Parameters

It is reported that about60percent controllers used in practicalindustrial sites suffer ever degrading performance due to poor controllerparameter tuning and maintenance.Aiming at constructing an expert system performing effective PIDcontroller parameter tuning, an adapted data-driven based multi-attributedecision-making fuzzy reasoning method is explicitly introduced alongwith a kind of fuzzy knowledge representation approach. In this context,a controller could be tuned by means of fuzzy inference imitating humanexpertise. Fuzzy matching degrees concerning selection ofdecision-making schemes are calculated based on fuzzification ofcontrollers’ multi-attribute indices before the aggregated results in termsof adjustment of PID controller parameters are achieved. Experimentalresults show that the proposed method enjoys simplicity as well aseffectiveness.To meet the applicable demand for control performance assessment and controller tuning, an OPC based controller tuning expert system（CTES）is accordingly developed. Taking advantage of OPC interface,industrial process real-time data are online captured and monitored beforecontrollers’ multi-attribute performance indices are achieved andfuzzified. Thereafter, controller tuning strategies are derived by fuzzyinference engine. Meanwhile, the performance results are stored in thehistorical database so as to facilitate later reviewing and analysis ofhuman operators. In addition, a mechanism of updating and verifyingknowledge-base is available thanks to the well established knowledgebase management system.By providing decision supports, CTES could help human operatorsto improve control performance of practical plants, being of attractiveengineering applicability potential.