Enhancing direct torque control of doubly fed induction motor in electric vehicle using artificial neural networks

Abstract

Electric vehicles (EVs) represent clean transportation solutions that play a pivotal role in the energy transition. By contributing to the reduction of greenhouse gas emissions, they pave the way for a more sustainable future. However, to ensure the full effectiveness of this transition, EV propulsion systems must meet modern demands for performance and efficiency. These systems must not only be robust but also capable of delivering high performance. In this context, Direct Torque Control (DTC), while widely used, has certain limitations. Specifically, it generates significant torque ripples and sometimes exhibits insufficient dynamic response to load variations. To address these issues, this work proposes an intelligent strategy referred to as ANN-DTC, which replaces hysteresis comparators and the switching table with regulators based on artificial neural networks. Simulations conducted in Matlab/Simulink demonstrate that ANN-DTC reduces torque ripples from 201.46 Nm to 87.31 Nm, an improvement of 56.66%, and eliminates the speed overshoots observed with conventional DTC. These results highlight the effectiveness of this approach in enhancing the dynamic response and adaptability of EV propulsion systems.