Multiparameters Motors Optimization Using Ensemble Surrogate-Assisted Multiobjective Differential Evolution Algorithm

Permanent magnet synchronous motors (PMSMs) are extensively utilized in industrial automation, transportation, and various other sectors owing to their superior efficiency, stability, and dynamic capabilities. Optimizing motor parameters is crucial for improving motor performance. Nonetheless, this optimization process often involves numerous time-intensive simulations, categorizing it as a costly optimization challenge. To address the issues of high computational costs and low optimization efficiency in motor parameter optimization, this study proposes an ensemble surrogate-assisted multiobjective differential evolution (ESMODE) algorithm. Initially, a framework for an ensemble surrogate model is established to reduce computational expenses and improve prediction accuracy. Subsequently, a population evolution strategy based on neighborhood field dynamics is implemented as the optimizer to speed up convergence. Furthermore, a criterion for infill sampling is designed to identify promising candidate solutions for training the model. The effectiveness of ESMODE is evaluated through experiments on standard benchmark test suites and comparisons with state-of-the-art expensive multiobjective optimization algorithms. Results demonstrate that the proposed method delivers superior performance. Ultimately, the ESMODE algorithm is applied to the optimization design of PMSMs. Comparative analysis of motor performance preoptimization and postoptimization confirms that the algorithm addresses key challenges in motor parameter optimization.