Self-tuning of a fuzzy logic controller using a cell state space algorithm

A method of automatically fine-tuning the output function parameters of a fuzzy logic controller based on the cell mapping concept is presented. The method takes a computational approach to the analysis of phase-space-based information about the global behavior of the system. The values of the output function parameters are determined with cell-state-based optimal control algorithm. The input-output data in the table are used to fine-tune the output function parameters of the fuzzy logic controller through a novel application of a mean-square-error gradient estimation algorithm. The accuracy of the cell mapping is increased by using variable time step sizes for cell transitions. The resultant fuzzy logic controller implements a smooth, highly nonlinear control surface that requires a relatively small number of rules. The validity of the method is demonstrated for the time-optimal setpoint control of a DC motor