Fairness-Oriented Charging Station Location Optimization Driven by Deep Reinforcement Learning

Against the backdrop of growing global attention to sustainable development and environmental protection, electric vehicles (EVs) have become an important choice for green transportation due to their zero-emission advantage. In recent years, the number of EVs in Beijing has rapidly increased, leading to an urgent demand for charging infrastructure. However, the existing layout of charging stations is imbalanced, with an over-concentration of stations in central urban areas and insufficient coverage in peripheral regions, hindering the further promotion of electric vehicles. Existing studies mostly rely on static data and traditional optimization methods, which are unable to adapt to the dynamic changes in the city, necessitating innovative strategies to improve both fairness and efficiency in location planning. This study aims to optimize the layout of charging stations in Beijing, balancing overall coverage with regional fairness to support the green transportation system. First, multi-source data including POI, population, GDP, and traffic networks are integrated, and GIS technology is used to analyze spatial distribution characteristics. Geographic detectors are then used to quantify the weight of influencing factors. Next, a fairness-oriented maximum coverage model is constructed, and Deep Reinforcement Learning (DRL) is employed to solve the problem, evaluating its advantages over traditional methods. The results show that the charging station layout in Beijing exhibits a “dense center, sparse periphery” pattern, with schools, hospitals, and other public facilities being the main driving factors. The DRL approach outperforms traditional methods in terms of coverage rate, fairness, and optimization efficiency, and can balance benefits and fairness by adjusting the parameter $\lambda $ This study provides theoretical and practical guidance for the planning of EV charging infrastructure in Beijing.