Improvement in speed performance of an induction motor with sliding mode controller and ANN for DTC

Abstract

To further improve the dynamic speed control performance of an induction motor (IM) using a controller based on sliding mode control (SMC) strategy, the switching table for direct torque control (DTC) is realized using a feed forward artificial neural network (ANN). The proposed feed-forward ANN consists of three layers: input, hidden and output layer. The input layer consists of three neurons (sector of flux vector, electromagnetic torque error and stator flux error), the hidden layer consists of a number of neurons that can be determined by experiment to obtain good results. The output layer consists of three neurons (three signals of the converter Sa, Sb and Sc). Simulation results under MATLAB environment are presented and compared with classical DTC using an Integral-proportional (IP) controller to verify the proposed approach.