On global analysis and design of fuzzy control systems

This paper presents a feedback design methodology that uses the combination of cell mapping and genetic algorithms to produce a progressively better fuzzy controller. This method uses global information about the system´s phase space behavior to adapt the controller. In this method, controller dynamics are parameterized; then starting with initial estimates of the parameters, new estimates are computed iteratively to optimize system response. The use of cell mapping in evaluating the global behavior of the closed-loop system is addressed. Performance comparison among cell-map-based design methods are shown using a DC motor as a test system