For AC machine is time-variant, nonlinear, multivariable, strongly coupled, in order to reach high accuracy of real-time control of speed and torque, a kind of DTC speed controller was designed combining sliding mode variable structure and fuzzy neural network control (FNNC). Variable boundary layer method was applied to eliminate chattering on the sliding surface; FNNC set the width of boundary layer on real-time. An improvement BP algorithm was designed to adjust the parameters of FNNC on line. Neural network identifier and PI speed identifier of MRAS were used to replace traditional speed sensor to identify the speed. Simulation and experimental results of the system showed that the robustness against parameters changes of induction motor is improved, and the static and dynamic performance is desirable. The controller will be a feasible plan.
|