DTC-SVM Control of induction motor Fed by PWM inverter using neural Networks

Abstract

This thesis presents a novel control strategy for induction motors by integrating Artificial Neural Networks (ANNs) into a Direct Torque Control with Space Vector Modulation (DTC-SVM) system. Traditional DTC methods, while providing fast dynamic response, suffer from high torque and flux ripples and variable switching frequencies. The inclusion of SVM improves performance but still requires precise tuning and suffers from model dependency. This work proposes the replacement of conventional PI controllers with ANN-based regulators trained on simulation datasets to optimize control performance. The proposed ANN-enhanced DTC-SVM is implemented and validated in MATLAB/Simulink. Simulation results confirm improved torque and flux regulation, reduced ripple, enhanced response time, and robustness against parameter variations, making it a suitable control strategy for industrial motor drive applications.