Multi-Objective Optimization of Contactor’s Characteristics Based on RBF Neural Networks and Hybrid Method

Contactors are typical electromagnetic devices whose optimization is multi-objective optimization problems. Contactors' static and dynamic characteristics obtained by the finite-element method (FEM) are typical objectives. Due to the low efficiency of FEM, the intelligent optimization algorithms that require multiple computations cannot be applied well here. In this paper, a fast algorithm for characteristics of contactors based on radial basis function neural network approximate model is proposed. The orthogonal least squares, data pre-processing and subsection modeling methods are adopted here to make the model more accurate. Meanwhile, a hybrid multi-objective optimization algorithm which combines the differential evolution and the particle swarm optimization is also proposed to speed up convergence and obtain the limit optimal solutions. The feasibility of the above-mentioned methods has been verified by solving a multi-objective optimization problem of the contactor with solenoid structure.