Optimizing Energy Management Strategy for EV Wireless Charging Efficiency Using Proximal Policy Optimization

Ava George, Jialin Tang

Abstract:

Wireless Electric Road Systems (wERS) embed copper
transmission coils beneath roads to dynamically transfer energy
to electric vehicles (EVs) in motion. As wireless power transfer
efficiency varies nonlinearly with vehicle speed, determining an
optimal adaptive speed control can improve dynamic wireless
charging performance and reduce electric vehicle (EV) range
anxiety. This paper trained a hybrid experimental computational
framework to identify optimal EV speeds for charging. A
physical prototype was constructed to emulate coil-to-vehicle
energy transfer at discrete traversal speeds (6.09,7.62, and 10.16
mm/s). Experimental results derived a relationship between
traversal speed and higher net battery gain. To extend these
findings, a Proximal Policy Optimization (PPO) reinforcement
learning (RL) agent was trained with a custom dynamic wireless
power transfer environment (DWPTEnv) using a synthetic speedcharge dataset. The environment models coil geometry, battery
state-of-charge evolution, and speed-dependent energy transfer.
Training results demonstrate stable policy convergence and high
value-function variance (0.93). The reward-per-step substantially
exceeded a random-speed baseline. This framework of combining
empirical and RL-based results provides a basis for further
smart-road EV energy management systems and real-time
dynamic charging optimization.