Design of PSO (Particle Swarm Optimization) Based Self-adaptive Fuzzy Controller

摘要

A self-adaptive fuzzy control method based on PSO membership function distribution is proposed in the article. The width of the coverage domain of the membership function and the influence of the overlap degree of neighboring membership functions on system output are considered in the method to dynamically adjust the membership function distribution so as to realize self-adaptive fuzzy control. Actually, the design of the fuzzy controller is equivalent to the process of finding optimal solution in high-dimensional fuzzy set, membership functions and system behavior search space. Positions of common particles are used to represent the fuzzy set distribution in corresponding domain, and before fuzzy inference, particle swarm evolution and iteration are used to adjust set partition in order to ensure that the fuzzy controller has optimal control precision under different responses. Additionally, effectiveness and superiority of the self-adaptive fuzzy controller designed thereby are verified through the simulation research on nonlinear system.