A Hybrid DNN–PSO Control Framework for Electric Vehicle Motor Drives with Optimized Power Modulator Design

Abstract

Control algorithms that can consistently handle nonlinear motor behavior and accurately shifting operating conditions are necessary to achieve excellent performance and energy economy in electric vehicle propulsion systems. To improve the performance of traditional PID-based motor control systems, this study proposes a hybrid architecture-control that combines a deep neural network with particle swarm optimization. To improve energy efficiency, PSO is used to optimize the controller parameters, while a DNN is used to capture the nonlinear dynamic behavior of EV traction motors. The proposed DNN–PSO controller is implemented in MATLAB/Simulink. The effectiveness of the proposed DNN–PSO control is evaluated. Performance comparisons with classical PID and ANN-based controllers indicate substantial energy gains, reaching up to 1.9% losses, with a faster speed response, with a rise time of approximately 0.5 s and an overall efficiency improvement of up to 92.6%. In addition, the proposed control technique exhibits better disturbance rejection and greater transient performance stability in comparison with AC motor control. This contributes to the improved hybrid control framework as appropriate for high-performance and energy-efficient EV propulsion applications.