HIGHER PERFORMANCE ENHANCEMENT OF DIRECT TORQUE CONTROL BY USING ARTIFICIAL NEURAL NETWORKS FOR DOUBLY FED INDUCTION MOTOR

Recently Direct Torque Control is widely appreciated compared to other conventional control methods due to its numerous advantages, notably its speed and precision. However, despite its qualities, it often encounters torque ripples that limit its operational effectiveness. These variations can be attributed to the use of hysteresis comparators, leading to variable frequency operation and undesirable speed overshoots. To address these challenges and enhance overall motor control, this article introduces a new approach based on neural networks. Direct Torque Control method is specifically designed for Doubly Fed Induction Motors and utilizes an Artificial Neural Network. Unlike conventional methods, this approach eliminates the need for speed controllers, commutation tables, and hysteresis comparators, thus providing a more integrated and efficient solution. Simulations conducted in the Matlab/Simulink environment have demonstrated the significant advantages of this approach with a higher performance enhancement. Not only were torque ripples reduced, but speed overshoot was completely eliminated. Furthermore, significant reductions in Total Harmonic Distortion values of stator and rotor currents were achieved, indicating an overall improvement in system performance.