Load Prediction of Charging Stations in Fast and Slow Charging Modes

[Purposes] The existing charging station load prediction mainly uses traditional statistical simulation or situation analyzation of electric vehicles (EVs) for forecasting, but the applicability and accuracy are not high enough. In response to this situation, it is proposed to predict the load of charging stations according to fast and slow charging modes. [Methods] First, the factors that affect the number of various types of EV in the charging station was analyzed, and BP neural network was used to predict the number of EVs in the charging station. Second, according to the different social functions of each type of EV, the proportion of fast and slow charging demands for each type of EV at each time was analyzed. Finally, LSTM and attention mechanism were used to predict the load of the charging station. [Results] The experimental results show that the accuracy of this method is higher than that of other existing methods.