Performance enhancement of DC motor drive for electric vehicle application by using deep neural network

In order to address the drawbacks of conventional proportional-integral (PI) controllers, including their inability to effectively handle nonlinearity, parameter fluctuations, and external disturbances, this research proposes a deep learning (DL)-based controller for DC motor drives in electric vehicles (EVs). The development of a custom neural network (CNN) controller is contrasted with neural net fitting (NNF) and PI controllers. The CNN controller is implemented and evaluated under various failure conditions and speed fluctuations after the DC motor drive system has been modelled. The CNN controller reduces overshoot by 25%, settling time by 30%, and speed tracking accuracy by 20%, according to the results. It keeps the system stable during high-resistance faults with a torque variation of only 15 Nm, whereas the PI controller becomes unstable when the torque variation reaches 25 Nm. This research comes to the conclusion that DL-based CNN controllers notably improve EV motor driving performance and dependability.