Affiliations: [a] Graduate School of Information of Technology, Hokkaido University, Sapporo, Japan | [b] OMRON Corporation, Kyoto, Japan | [c] Graduate School of Engineering, Nagasaki University, Nagasaki, Japan
Correspondence:
[*]
Corresponding author: Shuli Yin, Graduate School of Information of Technology, Hokkaido University, Sapporo, Hokkaido 060-0814, Japan. E-mail: shuli.yin@polimi.it, shuliyin@em.ist.hokudai.ac.jp
Abstract: An automatic design optimization of a wireless power transfer system is performed using Monte Carlo tree search (MCTS). Several key factors, i.e., the compensation network, shapes and geometrical parameters of the coils are determined after searches, in order to achieve the high transfer efficiencies for coaxial aligned and misaligned cases. The improved Selection policy in MCTS, called first play urgency (FPU), is implemented for optimization. It proves to be a more promising technique for finding optimal solutions when the searching volume is limited, compared with the original MCTS with using the upper confidence bound applies to tree (UCB1).
Keywords: First play urgency, Monte Carlo tree search, optimal design, wireless power transfer systems
