Design of sliding mode model predictive dual-loop control through self-learning strategy to mitigate the torque ripple in BLDC motor for electric vehicles

This study investigates a novel dual-loop control strategy that combines sliding mode and model predictive controllers to reduce torque ripple in high-performance Brushless Direct Current (BLDC) motors, especially for automotive electric vehicle (EV) applications. The proposed control system merges the predictive features of Model Predictive Control (MPC) with the robustness of Sliding Mode Control (SMC), creating a dual-loop structure that optimizes inner-loop current regulation and outer-loop speed control. The cost function is formulated to regulate the d- and q-axis currents, enabling the calculation of the optimal output voltage signal necessary for efficient motor performance. This synergy ensures precise stator current modulation, effectively reducing torque ripple while maintaining superior motor efficiency and stability. Additionally, by incorporating adaptive heuristics and data-driven insights through a hybrid self-learning algorithm combining ANN and fuzzy logic, the SMC-MPC controller can forecast and reduce error rates in the BLDC motor, ensuring smooth torque output with minimal ripple. The performance of the SMC-MPC strategy is thoroughly evaluated through MATLAB/SIMULINK Model-in-the-Loop (MIL) simulations and validated via Hardware-in-the-Loop (HIL) testing. Comparative analysis shows that the proposed controller provides superior results, including a rapid 0.01 s rise time, a minimal 0.001% steady-state error, a 0.02 s settling time, and a peak overshoot of 0.066%, outperforming traditional PID and SMC controllers. Also, the experiments show a 28.57% reduction in torque ripple and efficiency maps, achieving 96.47% maximum efficiency. This endeavor validates that the SMC-MPC controller improves BLDC motor efficiency while extending the operational range of EVs.