A Bayesian Optimization and Partial Element Equivalent Circuit Approach to Coil Design in Inductive Power Transfer Systems

Abstract

The design of transmitter and receiver coils is essential for Inductive Power Transfer (IPT) systems. However, traditional coil design depends on trial-and-error methods. These methods usually require a great number of finite-element simulations, which is time consuming. This paper adopts a Bayesian Optimization and Partial Element Equivalent Circuit (BO-PEEC) approach for coil design. The Bayesian Optimization (BO) algorithm is used to reduce the number of simulations. Partial Element Equivalent Circuit (PEEC) theory is applied for coil modeling to reduce evaluation time in each simulation. Compared with traditional parametric-sweep-and-finite-element-simulation procedure, the proposed BO-PEEC framework can significantly accelerate the design speed and achieve accurate design. This is verified by both simulations and measurements.