With the gradual promotion of electric vehicles, research on the charging load characteristics of electric vehicles is not only conducive to the optimal operation of the charging station, but also conducive to the safe and stable operation of the power system. According to the time and space characteristics of the electric vehicle charging load, this paper proposes a research method based on an autoencoder. Based on the analysis of the NHTS data set, the time distribution law of the electric vehicle charging load is determined. The characteristics of electric vehicle travel mileage and travel end time are extracted by an autoencoder method. On this basis, the probability of an electric vehicle reaching different destinations is calculated, and the time-space characteristic model of electric vehicle charging load is established. The shortest waiting time of each electric vehicle is calculated, and the time error is considered when calculating the load. This improves the accuracy of the calculation of charging load. Finally, the charging load of electric vehicles in the region is obtained by an example which verifies the feasibility and accuracy of the proposed research method. This work is supported by the National Natural Science Foundation of China (No. 61672337).