The Research Of Adaptive Robust Fuzzy Control Methods For Nonlinear Systems

In adaptive fuzzy control researches, the direct and indirect adaptive fuzzy control approaches are usually used in many schemes which just utilize the knowledge of control behavior or plant. But the knowledge of human experience or experts can not be used sufficiently in these control schemes and most methods are designed under the assumption that approximation errors satisfy squared integral conditions. Furthermore, adaptive laws presented in most schemes all adjust optimal approximation parameter vectors. However, in practice, the approximation error is difficult to satisfy squared integral conditions, and online computational burden is very heavy if the estimate of optimal approximation parameter vector is adjusted adaptively.Based on the above problems, this paper presents two kinds of control methods:1) For a class of nonlinear systems with disturbances, a combined adaptive fuzzy control method with the function of continuous supervisory control is presented in this paper. This method utilizes the knowledge of plant and control behavior adequately. The controller is composed of three parts: tracking control part, continuous supervisor part and adaptive compensator part. The tracking control part is used to obtain desired output tracking performance; the continuous supervisor part is to guarantee the boundedness of system states, and the adaptive compensator part is to simultaneously compensate the approximation error of fuzzy logic system and the effect of the external disturbances. In stability analysis, the term of minimum approximation error to be squared integral is not needed. Simulation results show that better tracking performance is obtained by using this method compared with the method only using the knowledge of control behavior or plant.2) An adaptive robust fuzzy control method is proposed for a class of uncertain MIMO nonlinear systems with disturbances. The effect that function approximation error and the external disturbances have on the tracking error is compensated by means of the robust controller designed. It is not required to assume that estimation error has a known boundary or satisfies squared integral conditions and even it is not required that the states of the system are full observable. It is just supposed that the boundary of estimation error and external disturbance are unknown. The online computational burden is greatly reduced since only uncertainty bounds are tuned online.