At present, most of the research on electric vehicle load forecasting does not focus on the actual area. In this paper, data driven and model driven are combined to predict the load of electric vehicles. On this basis, an automatic demand response strategy is proposed. Firstly, through the data collection of Didi, the regeneration information such as functional area division, electric vehicle travel time and space transfer matrix are obtained. Considering multi-day charging, the load forecasting model is constructed. Secondly, in order to stabilize the fluctuation of net load, considering the benefits of both sides of supply and demand, the automatic demand response strategy is formulated. Combined with the benefits of price type and incentive type demand response mechanism, the uncertainty of user participation is described by introducing logistic function. Finally, taking a certain area in Haikou as an example, the simulation results verify that the prediction model can effectively predict the load of the actual area, and the proposed strategy can achieve a win-win situation between supply and demand, consume absorb new energy locally, and stabilize the load fluctuation.