Smart Management of Electric Vehicle Chargers Through Reinforcement Learning

A reinforcement learning (RL) based EV charging management system is developed for the charger coordination problem. RL can handle system uncertainties, requires no historical data, and is not affected by future changes to the system. The EV charging coordination reduced peak demand, required charging infrastructure and increased solar power utilization. The proximal policy optimization (PPO) is used to train the RL to reach targets and the results are compared with other benchmark algorithms.