User Cost Minimization and Load Balancing for Multiple Electric Vehicle Charging Stations Based on Deep Reinforcement Learning

In the context of global energy conservation and emission reduction, electric vehicles (EVs) are essential for low-carbon transport. However, their rapid growth challenges power grids with load imbalances across networks and increases user charging costs. To address the issues of load balancing across large-scale distribution networks and the charging costs for users, this paper proposes an optimization strategy for EV charging behavior based on deep reinforcement learning (DRL). The strategy aims to minimize user charging costs while achieving load balancing across distribution networks. Specifically, the strategy divides the charging process into two stages: charging station selection and in-station charging scheduling. In the first stage, a Load Balancing Matching Strategy (LBMS) is employed to assist users in selecting a charging station. In the second stage, we use the DRL algorithm. In the DRL algorithm, we design a novel reward function that enables charging stations to meet user charging demands while minimizing user charging costs and reducing the load gap among distribution networks. Case study results demonstrate the effectiveness of the proposed strategy in a multi-distribution network environment. Moreover, even when faced with varying levels of EV user participation, the strategy continues to demonstrate strong performance.