With the continuous promotion of electric vehicles and the construction of charging facilities, electric vehicles will be more and more applied, and the probability density prediction of charging load of electric vehicles is the basis of the construction and operation of charging stations. Therefore, this paper proposes a probabilistic model prediction method for charging load of Electric Vehicles (EV) considering different spatio-temporal travel characteristics of users. Firstly, user behavior characteristics are analyzed, including user travel chain structure, user travel time, battery usage, etc. Secondly, the optimal travel chain is obtained by Using Dijkstra path searching algorithm. Finally, by using Bass model, Monte Carlo method and nonparametric kernel density estimation method, the probability density function of electric vehicle charging load under different scenarios is obtained, providing data support for the planning, design and operation of charging stations.