Model of Load Forecasting of Electric Vehicle Charging Station Based on SAE-ELM

The randomness of charging behavior in time and space increases the difficulty of load forecasting of EV charging station. In this paper, the stacked auto encoder neural network-extreme learning machine (SAE-ELM) hybrid model is proposed by improving the stack auto encoder of deep-learning to realize short-term load forecasting of charging stations. The interactive mode of electric vehicle and power grid is introduced and the key factors affecting the charging station load, such as historical load, environment, typical day type, etc., are also considered. Finally, the short-term load forecasting of a practical charging station is realized and compared with SAE-BP and ELM algorithm. The result shows the proposed approach can provide more accurate forecasting result, which benefits the stable operation of power grid.