Multi-objective optimization algorithm based on electromagnetic relay parameters

Electromagnetic relays perform critical functions such as control, circuit switching, and safety grounding within power equipment and systems, fulfilling roles that cannot be replaced by semiconductor switches. This paper addresses the issue of low qualification rates caused by the two-step absorption and release processes in balanced force electromagnetic relays. A multi-objective optimization study is conducted on these relays, focusing on the structural parameters of the electromagnetic system. The Non-dominated Sorting Genetic Algorithm III (NSGA-III) is employed to solve the Pareto frontier solution of multi-objective higher-order functions due to its high convergence speed, search efficiency, robustness, and portability . This paper proposes a novel multi-objective optimization algorithm based on NSGA-III, utilizing Isolation Forest (iForest) and Support Vector Regression (SVR) for the structural parameter optimization of electromagnetic systems in balanced force relays. Firstly, the original data are processed using the iForest method to detect and eliminate outliers. Secondly, given the small sample s ize of the electromagnetic suction force dataset for balanced force relays, the SVR model is used to predict and expand th e dataset, thereby providing a solid data foundation for subsequent structural parameter optimization. Thirdly, constraint and objective functions are constructed, and the NSGA-III algorithm is used to predict the electromagnetic suction force through the SVR model, optimizing the structural parameters. The NSGA-III algorithm refines the structural parameters of the electromagnetic relay using data generated by the SVR model to obtain the optimal parameters. Finally, finite element simulation software verifies the results, demonstrating that the proposed method effectively identifies the optimal parameters of the electromagnetic rela y and enhances optimization efficiency.