Two-Step Monte Carlo Tree Search for Optimal Design of High-Frequency Toroidal Inductors in Power Electronics Circuits

Abstract

This study devised a novel optimization technique for toroidal inductors utilizing a two-step Monte Carlo tree search (MCTS) algorithm. The proposed method integrates a global search based on analytical formulas with a local search involving 3-D finite element analysis (FEA). By employing this approach, the optimal core materials, sizes, number of turns, and windings are determined to enhance efficiency and reduce the overall size of the inductors. The proposed method was applied to optimize inductors in a 6.78 MHz electrically coupled power transfer system. The results of the study demonstrate that the optimized inductor, identified through the proposed method, not only achieved superior efficiency but also minimized computational costs. Subsequent system measurements confirmed that the optimized inductor improved system efficiency by over 1.8% compared with inductors designed through empirical methods.