Abstract: The paper proposes a partly continuous fuzzy algorithm that is combined with the existing method to improve the real time ability and the precision of fuzzy controller obviously. A time-varying modifying factor fuzzy controller is constructed by the method. Then a novel simulated annealing algorithm is presented to rapidly achieve the optimal result. The self-optimizing fuzzy controller is made of the two algorithms. The simulation results illustrate that the self-optimizing fuzzy controller has fine performances for dead-time plants and non-stable nonlinear plants whose mathematical models are not known. It is the main advantage that the self-optimizing fuzzy controller is independent of human experience and the mathematical models of plants. The paper tries to provide a way for controlling non-linear plants of unknown mathematical models.