Design of a novel fuzzy neural controller in DTC system

Abstract

Induction motor is a nonlinear and time-varying dynamics system. In order to improve performance of direct torque control system, a fuzzy neural controller that is not depend on the exact object model is applied. The conventional fuzzy neural control algorithms have double inputs and complex computation, and suffer from high computational burden. A novel fuzzy neural controller with single input, low computation, five-layer structure is presented for direct torque control system of induction motor. The performance of the novel controller is same to the conventional fuzzy neural network controller, better than PI controller, and it reduce requirement of the microcontroller operation speed because of little input data quantity and simple computation. The simulation and experiment results show that, compared the conventional PI controller, the simple fuzzy neural controller reduces overshoot and settling time of the speed step response, improves the control performance of direct torque control system. The method provides an optional ways for the design of high performance and low cost industrial driver controller.