Defuzzification Filters and Applications to Power System Stabilization Problems

Defuzzification is a very important step in fuzzy systems applications. There are a number of different defuzzification methods reported in the literature. In this paper, the concept of defuzzification filters in a control system setting is first discussed and a methodology for designing such filters considered. As will be seen, the design of such filters requires the knowledge of the plant model and its inverse. A reference control signal is computed and then is used to generate the actual defuzzified control signal which will be applied to control the plant. The application of the defuzzification filter is made by introducing the filter into a power system in which a neuro-fuzzy self-learning controller was applied to stabilize the system but success could not always be guaranteed. With the defuzzification filter, however, the system is always stabilized. Simulation results are presented