Simulation Model of Direct Torque Control for Induction Motor Based on Artificial Neural Networks


Abstract: Direct torque control became the most popular technique for induction motor control through the last two decades, because of its simple structure, accurate and fast torque response, but it has some drawbacks such as torque and stator flux ripples. Therefore, an accurate and fast estimation of stator flux and torque values is required. In this paper a proposed model for two Multi-layer Feed-Forward Neural Network (MFFNN) to simulate and train the direct torque control data of three phase induction motor for estimation of electromagnetic torque, stator flux, and flux angle at two different sampling frequencies. The feed-forward neural networks proposed consist of three layers. The input layer consists of four neurons (stator voltages and currents) and the output layer consists of three neurons (electromagnetic torque, stator flux and flux angle). Quick back-propagation algorithm is used to train the proposed networks. Simulation model is performed using MATLAB. The results have been compared according to computation time and accuracy.