A Noble Fuzzy Self-Tuning Scheme for Conventional PI Controller

In this study, we propose an improved auto-tuning PI controller. The widely practiced Ziegler-Nichols tuned PI controller (ZNPIC) exhibits poor performance for nonlinear and high-order systems. In order to have an overall improved transient response, the proportional constant of the proposed PI controller is parameterized by a real-time nonlinear updating factor ( depending on the process trend. The value of ( is determined through a fuzzy inference engine with 49 if-then rules defined on the process error and change of error. Performance of the proposed fuzzy self-tuning ZNPIC, termed as FST-ZNPIC, is tested for second-order linear and nonlinear dead time processes including a marginally stable system under both set point change and load disturbance. Simulation results reveal our FST-ZNPIC provides significantly improved transient responses compared to other conventional and fuzzy self-tuning controllers. Robustness of FST-ZNPIC is observed with a considerable change in process dead time.