Self-tuning Fuzzy-PID controller design

A self-tuning Fuzzy-PID structure is proposed as a new controller model. While it maintains the desirable characteristics of PID control, the Fuzzy-PID controller provides additional flexibility with its nonlinear gain characteristic and adaptive operation. Although the PID controller is by far the most common control algorithm and widely used in many industries, it is often poorly tuned, and the constant gain structure limits its control capability. Fuzzy control (FC) has received much attention and can be found in many applications, but the stability and robustness aspects of FC are not well-understood. FC has not been utilized in many real-time applications where safety is a key requirement. A new approach that merges these two control algorithms is proposed to take advantage of both the well developed PID structure and the potential of fuzzy knowledge representation. The control signal is generated by an input-dependent PID controller whose gain parameters are obtained from inference on a fuzzy rulebase. The rulebase can be adapted to different plants through successive refinement learning. Properly designed Fuzzy-PID controllers can achieve better performance and have greater control design flexibility than traditional controllers. An analysis of the FuzzyPID controller's stability and robustness properties in a closed-loop setting is provided.