Structure Analysis Of Typical Fuzzy Controllers And Study Of Interpolation Mechanism

Structure analysis of typical fuzzy controller can offer important guidance for designing fuzzy controller. The fundamental aspects of fuzzy controller systems, such as interval transforms and inference methods, have obvious influence to the control performance.Dead-zone often exists in fuzzy control system especially when trapezoid membership functions were used. Based on Mamdani's sum-min, max-min, max-product inference operator, the structure characteristics of the three classes of fuzzy controllers are investigated and compared. The different characteristics of fuzzy controllers using different membership functions are also studied. Finally, the internal relation between PID controller and fuzzy controllers are also discussed. These provide some theoretic bases for designing fuzzy control systems.The equivalence between the Mamdani type fuzzy systems using trapezoid or gauss membership functions and the piecewise interpolation functions is demonstrated. And the estimation of the approximation error is given. Computer simulation results make clear that fuzzy controller using interpolation arithmetic has better control performance than the traditional one because it can eliminate quantization dead-zone and adjust steady-state errors completely. Compared with PID controller, it suits for the systems with changeful parameters and structures much better than PID controller.