Modeling and implementation of a machine learning-based wireless charging system with high misalignment tolerance

Abstract

Wireless charging for electric vehicles has an alignment problem that affects transmission efficiency. Studies aimed to reduce alignment issues by employing techniques such as varying coil structures, altering the charging system topology or frequency, and implementing phase shift control. A new work allowed for an alignment tolerance of 220 cm on the x-axis, 110 cm on the y-axis, and 30 cm on the z-axis. Eight scenarios were analysed for stable solutions, mutual inductance (M) and transmission power of electric vehicles. The study resolved the misalignment and increased transmission power by 28.8 %, 68.7 %, and 67.8 % for Scenarios 1–3 compared to the unaligned situation. In Scenarios 4–7, the proposed system can transfer power despite being in the unaligned state. The power transfers are 983 W, 968 W, 1006 W and 971 W, respectively. In Scenario 8, power transmission is impossible with or without the proposed system because of the long distance between coils.

Keywords

• High tolerance misalignment
• EV charging
• Machine learning