Optimization of Magnetic Couplers in Wireless Charging Systems Using the Surrogate Model

Abstract

This article proposes a general and efficient surrogate model-based parameter optimization method, consisting of the Taguchi method, finite element method (FEM) simulation, genetic algorithm-backpropagation (GA-BP) neural network, and grid search, to optimize various magnetic coupler with multiple design objectives in wireless charging systems (WCSs). First, the WCS with hub-type multiple circular coil coupler (HMTC) is used as an example to validate the necessity of parameter optimization for the magnetic coupler. Then, the HMTC’s structure and principles are analyzed. Three design objectives with function expressions are used as examples for designing an HMTC: the rated mutual inductance, high misalignment tolerance, and lightweight receiver. Second, a detailed workflow of the proposed method is analyzed, especially the basic working principles of the Taguchi method and GA-BP neural network-based surrogate model. Third, the HMTC optimization process and the proposed method’s main parameter determination are analyzed. Then, simulation results verify the proposed method’s feasibility on HMTC and two other magnetic couplers. Numerical results prove that the proposed method has higher optimization efficiency and accuracy. Finally, a prototype is built. Experimental results confirm that the rated mutual inductance is $20.2~\mu $ H, and its fluctuation within ±5% is circular (radius is 34% of the transmitter diameter).