ABSTRACT The reduction of torque ripple in PMSM (Permanent Magnet Synchronous Motors) is not only a major factor in the performance and comfort of electric vehicles (EVs) but also a necessity. Torque ripple is the greatest challenge in PMSM, causing noise, vibration, and lower‐efficiency in EVs. Here, this paper introduces a hybrid method to reduce torque ripple in PMSM applied to EVs. The developed hybrid technique is the simultaneous execution of both the Improved Dwarf Mongoose Optimisation (IDMO) and the Supervised Attention Multi‐Scale Temporal Convolutional Network (SAMTCN). Therefore, it is referred to as the IDMO‐SAMTCN technique. Its objective is to enhance motor performance, increase driving comfort, and improve vehicle dynamics overall. The IDMO technique is applied to gain the optimal value of the PI controller. The SAMTCN technique is used to predict the optimised parameters. The combination of IDMO and SAMTCN enhances the advantage of accurate prediction and reduction of torque ripple, which leads to reduced noise and extended durability of the EV parts. The next step was to create the method in MATLAB. Compared to existing optimisation algorithms like PSO (Particle Swarm Optimisation), ACO (Ant Colony Optimisation), and Genetic Algorithm (GA), the IDMO‐SAMTCN method stands out as the most efficient model. The IDMO‐SAMTCN method reaches an impressive 95.9% efficiency, TWHD, and torque ripple of 4.87% and 6.83%, respectively, the lowest in comparison with existing techniques.
Ravichandran et al. (Thu,) studied this question.