Integral-action predictive speed control with neural network current regulation for PMSM electric vehicle drives: Co-simulation hardware-in-the-loop validation

As transportation systems shift toward electrification and decarbonization, developing robust, high-performance control strategies for electric vehicle traction systems has emerged as a key technological priority. A novel hybrid robust intelligent control approach is proposed in this study, integrating an artificial neural network (ANN) within a robust predictive speed control framework for permanent magnet synchronous motor (PMSM) based electric vehicle propulsion systems. The nonlinear mapping capability of the ANN, driven by sigmoidal activation functions, enables the controller to effectively approximate complex system dynamics without requiring explicit disturbance estimation. This adaptive feature enhances compensation for modeling inaccuracies and parametric uncertainties, thereby improving tracking precision and significantly reducing oscillatory behavior. Furthermore, the speed control loop is formulated through a newly designed predictive cost function that incorporates integral action into the optimization process, ensuring enhanced disturbance rejection capability and superior dynamic performance under parameter uncertainties and external perturbations. The effectiveness of the proposed control strategy is validated through hardware-in-the-loop (HIL) experiments that integrate an OPAL-RT simulator, which runs the system components, with a dSPACE-based controller that executes the control. The results confirm the effectiveness of the proposed RPNN strategy in improving system performance. Compared with the conventional control, the proposed method significantly reduces torque ripple by 86.95% and decreases the current THD by 79.12%. In terms of speed control, the response time is considerably shortened, achieving an improvement of 98.79%, while the disturbance rejection time is reduced by 87.5%. These results demonstrate the superior dynamic performance and robustness of the proposed control approach.