Research on multi-objective optimization of permanent mag-net brushless dc motors based on a grouped surrogate model with parameter correlation

To mitigate cogging torque and torque ripple while enhancing electromagnetic torque in permanent magnet brushless DC motors (PMBLDCM), this paper explores a multi-objective optimization strategy utilizing a parameter correlation-based grouped surrogate model. Initially, an orthogonal experimental design integrated with Spearman's rank correlation coefficient is utilized to analyze, sift, and categorize the stator and rotor parameters. Subsequently, a holistic optimization strategy leveraging the Kriging surrogate model, a multi-objective particle swarm optimization algorithm, and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) is implemented for global optimization of the highly correlated parameters. Next, for parameters with low correlation, the entropy weight technique is applied to assign objective weights, and a polynomial response surface surrogate model is developed, followed by local optimization via a nonlinear sequential quadratic programming algorithm with gradient-based optimization. Ultimately, a comparative evaluation of the electromagnetic performance is performed, contrasting the optimized motor against the original design. The outcomes demonstrate a 7.6% enhancement in electromagnetic torque following the multi-objective optimization with the parameter correlation-based grouped surrogate model, alongside a 23% and 16% reduction in torque ripple and cogging torque.