An Optimization Strategy Based on Dimension Reduction Method in Wireless Charging System Design

This study proposes an optimization strategy, which is based on a dimension reduction method with sensitivity analysis, for fast multi-objective optimization of coil design in the wireless power transmission (WPT) system. The multi-objective optimization process of coil in WPT usually involves general genetic algorithm (GA) and derivatives that suffer from common problems of slow search speed and heavy calculation burden, especially in the presence of multiple optimization variables. Moreover, in a complicated WPT system with ferrite, the optimization objective, such as the coupling factor
$k$
, is difficult to calculate analytically; in practice, it is acquired numerically using the finite element method (FEM), which also adds to the optimization complexity and time consumption. As a remedy, we propose the new strategy by simplifying the original complex optimization problem firstly with sensitivity assessment, and then applying NSGA-II convergence algorithm for optimization. Firstly, the optimization strategy is introduced, including the evaluation method, dimension reduction method and entire multi-objective optimization process. Then the optimization strategy is implemented to optimize the design parameters of a coil system. The sensitivity of design parameters is assessed with FEM analysis to realize the dimension reduction, and the optimization framework is established. Afterwards, the high sensitivity parameters will be optimized utilizing the multi-objective algorithm to obtain the optimal parameters. The simulation results demonstrate that, compared to the traditional NSGA-II algorithm, the proposed strategy yields a 44% reduction in the iteration time required for system convergence. Furthermore, the good agreement between the simulation and the measured sensitivity data and optimal parameters presents additional validation for the optimization strategy. The experiment further shows that implementation of the proposed strategy could lead to an increase in the maximum efficiency of the WPT system from 79.85% to 93.2%.