Abstract: The charging load of electric vehicles is random in the spatial and temporal distribution, and the road network structure and traffic congestion are important factors affecting the spatial and temporal distribution of the load. The traditional road network structure model often ignores the grade and bending characteristics of the road and cannot better reflect the time-sharing congestion of each road section. Firstly, the road data attributes are divided by ArcGIS. Python processes the real-time road condition layer of the Amap Open Platform to obtain the weighted congestion coefficient of each road section in different periods to construct the road network model expression. Secondly, the distribution characteristics of urban points of interest in the study area are analyzed, and the urban functional areas are divided by the kernel density estimation method. Then, the travel chain model is used to analyze the travel behavior characteristics of electric vehicle users to improve the Floyd algorithm to select the shortest travel route. The improved Floyd algorithm solves the redundancy calculation problem of the traditional Floyd algorithm, reduces the algorithm's complexity, and predicts the charging load of electric vehicle users in the region through the Monte Carlo method. Finally, the effectiveness of the proposed method was verified by simulating and predicting the charging load demand of electric private cars in a certain district of Lanzhou within a day. The results show that the proposed method can more intuitively reflect the distribution characteristics of charging load demand in different functional areas of the region, improving the overall load forecasting accuracy.