Aiming at the location and capacity determination of charging piles for electric vehicles in urban areas, considering the demand forecasting of charging load, a location and capacity determination model was established by bi-level programming. Firstly, according to the demand forecasting of charging load, the load forecasting model was established by using autoregressive model. Taking the node and capacity of charging pile planning as control variables and the minimum cost of system comprehensive planning as objective function, the upper model was established based on network topology and charging load forecasting, taking into account the necessary constraints such as equipment capacity allocation constraints. Taking the charging scheme of electric vehicle users as the control variable, the lower model was established to minimize the comprehensive charging cost of users. The lower model makes charging plan based on the location and capacity scheme of the upper model, and feeds back the comprehensive charging cost to the upper model. The upper model needs to take the strategy of the lower model into account. The model solving process was designed by using firefly algorithm. Finally, an example was given to optimize the location and capacity of the charging station for electric vehicles in a new area of Xi"an city. The validity and correctness of the model are verified.