Load Analysis and Prediction of Electric Vehicle Charging Stations Supported by RF-CNN Algorithm

Abstract

Aiming at the problems of low prediction accuracy and single data type in traditional electric vehicle charging station load prediction models, a RF-CNN algorithm based on the combination of random forest algorithm and convolutional neural network is proposed to improve the accuracy and efficiency of electric vehicle charging station load prediction. The research first takes the state of charge of electric vehicles and charging stations as the research object to construct a load prediction model. Then it utilizes the designed algorithm to optimize model performance. Finally, different validation indicators are used to verify the predictive performance of the prediction model in the load of charging stations. Finally, the prediction accuracy of the model in the charging station load prediction reaches 92.71%, which is higher than 85.16% of BP and 88.09% of LSTM. The loss value during training was 11.34%, lower than 38.06% for BP and 28.52% for LSTM. The average percentage error, average response speed and data generalization ability were 1.82%, 42ms and 83.61%, respectively, which were significantly better than the comparison method. This indicates that the model has higher accuracy and robustness in load forecasting of electric vehicle charging stations, and could improve the operational efficiency of the power system. It has important application value in load analysis and prediction of electric vehicle charging stations.