Abstract: The layout of electric vehicle (EV) charging stations directly affects the investment and construction cost of operators, and it also relates to the convenience and economy of users' travel. In order to balance the benefits of charging station operators and users, and provide reasonable solutions for configuring charging piles in a station, this paper studied the planning method of EV charging stations. Firstly, the mathematical model of EV charging station location selection and capacity configuration was proposed to minimize the cost of charging station and the economic loss of users (including time loss and power loss). Then, based on the historical data of EV population, BP neural network was applied to predict the future time distribution of EV in the target planning area. In addition, the actual traffic data of the target area was used to determine the spatial distribution of EV. Finally, the dynamic probability mutation method based on regional visits was proposed to improve the particle swarm optimization algorithm. The algorithm was used to solve the EV charging station location selection and capacity configuration model, and the charging station planning optimization scheme under the current residents' income level was obtained. Taking Haidian District of Beijing as an example, the case analysis showed that the optimization model proposed in this paper had the advantages on economy and user satisfaction. Meanwhile, the improved particle swarm optimization algorithm had fast optimization speed and strong adaptability.