An Online Data-Driven Multi-Objective Optimization of a Permanent Magnet Linear Synchronous Motor
Xiao Liu, Chunfu Hu, Xiongsong Li, Jian Gao, Shoudao Huang
Abstract
In this article, an online data-driven multi-objective optimization is proposed to suppress the undesired thrust ripple and improve the average thrust of the permanent magnet linear synchronous motor (PMLSM). A two-loop optimization process is adopted, and the first loop is designed to achieve a higher accuracy of the Kriging models by adding points based on expected improvement (EI) criteria. Through the first loop, the Kriging models achieve a significant increase in the prediction accuracy from 76% to 99.7%. Then, the second loop runs to optimize the PMLSM by using the Kriging assisted multi-objective particle swarm optimization (MOPSO) method. The results show that the optimized PMLSM successfully achieve a 64% reduction in the thrust ripple and a 6.6% increase in the average thrust compared with the original PMLSM.