Asian Engineering Review

This study proposes an intelligent control framework that integrates deep reinforcement learning (DRL) and multi-objective particle swarm optimization (MOPSO) to enhance the dynamic performance, energy efficiency, and robustness of three-phase asynchronous motor drives applied in sustainable electrification systems. The proposed framework integrates a DRL agent with an MOPSO optimizer within an adaptive control architecture. The DRL controller learns optimal control policies through continuous interaction with the motor drive system, while MOPSO simultaneously optimizes multiple conflicting objectives, including torque ripple minimization, total harmonic distortion (THD) reduction, speed-tracking accuracy improvement, and energy-efficiency enhancement. The controller is evaluated using a high-fidelity MATLAB/Simulink induction motor model under load disturbances, parameter variations, and stochastic uncertainties. Simulation results demonstrate torque ripple reductions of 25–30% compared with conventional Field-Oriented Control (FOC) and Direct Torque Control (DTC). THD decreases from 6.8% (DTC) and 4.1% (FOC) to approximately 1.9%, while speed-response settling time is reduced to 0.12 s with less than 4% overshoot. Drive efficiency improves by 8–12%, and performance deviations remain below 5% under uncertain operating conditions. The proposed DRL–MOPSO framework significantly outperforms conventional control methods in terms of efficiency, dynamic response, and robustness. The framework offers a viable intelligent control solution for electric vehicles, industrial automation, and renewable-energy-powered drive systems, supporting sustainable and reliable electrified applications.