A New Fuzzy Gain Scheduling Algorithm for Process Control

A new fuzzy gain scheduling technique is proposed that involves implementing several linear controllers over a partitioned process space. A higher-level rule-based controller determines which local controller is executed. Unlike conventional gain scheduling, a controller with fuzzy gain scheduling uses fuzzy logic to dynamically interpolate controller parameters around region boundaries based on known local controller parameters. The tuning of the controller is band on local models and thus overcomes the usual problem of tuning the fuzzy logic controller. Simulation of various gain-scheduled PID controllers for a nonlinear plant demonstrate that controllers with fuzzy schedules out-perform those with conventional schedules and are comparable to adaptively-tuned PID controllers. Based on the analysis of some tuning rules and other rules of thumb, new fuzzy gain scheduling rule are formulated that enhance the performance of PID controllers.