Charging Control of an Electric Vehicle Battery Based on Reinforcement Learning

摘要

Electric vehicle (EV) charging has started to attract people's interest due to the booming development of EVs. However, uncontrolled charging of EVs may increase users' charging cost, considering an hourly-changing electricity price. Existing methods based on solving optimization problems place a high demand on the accuracy of a given battery model, which is difficult to acquire due to the limited types of battery models and parametric uncertainties. Therefore, it is necessary to discover a novel way to address this problem without a precise mathematical battery model. Therefore, we propose a charging control method based on reinforcement learning (RL) to seek an optimal charging portfolio to minimize charging costs, which can be battery model free and data driven. Moreover, the presented control algorithm provides a basic framework for a more complicated electricity market where various types of generators and loads exist.