A Data-Driven Approach for Real-Time Residential EV Charging Management

When electric vehicle (EV) participates in demand response with real-time electricity price, the EV charging cost can be significantly reduced by properly managing the charging schedules according to these pricing data. However, due to the existence of randomness in the pricing process of the utility and the user's commuting behavior, determining a cost-efficient charging strategy becomes challenging. Traditional model-based solutions need a model to predict the uncertainty. Constructing a model-based controller is difficult when the heterogeneity of EV users is taken into consideration. In this paper, the EV charging management problem is formulated as an Markov Decision Process (MDP) which has unknown transition probability. A data-driven approach based on deep reinforcement learning is developed to determine the optimal EV charging strategy. The proposed approach does not need any system model information. Experimental results verify the effectiveness of our proposed approach.