Surrogate-Assisted Multiobjective Optimization of Double-D Coil for Inductive Power Transfer System With <i>LCC</i>–<i>LCC</i> Compensation Network
Yadong Wang, F. Wang, Ye Tian, Aoni Sun, Bangyin Liu
Abstract
Machine-learning algorithms have been widely researched in the inductive power transfer system to find optimal coil geometry. However, this method requires a large amount of training samples, and it is difficult to reach an optimum design if there are many design criteria. A surrogate-assisted multiobjective optimization method considering compensation parameters is proposed and implemented for double-D coils with <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LCC</i> – <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LCC</i> topology, which can quickly give the Pareto front of coupling coefficient, volume, and stray field with few finite-element method simulations. It is achieved by combining the Taguchi method and extreme learning machine (ELM) training. The system configuration, optimization objectives, and design variables are first analyzed. Then, Taguchi method and ELM theory are presented in detail. The multiobjective design process and optimization results are further demonstrated. Finally, a 2.5-kW hardware topology is constructed and a peak efficiency of 96.5% is achieved. The experimental results verify the correctness of the theoretical analysis and the effectiveness of the proposed method.