Electric vehicle charging load forecasting based on GA-CNN-GRU model

The inherent randomness of electric vehicle (EV) charging habits significantly complicates the task of accurately predicting the immediate energy requirements at EV charging stations. Addressing the low accuracy of short-term forecasting of charging loads at charging stations under the interference of multiple influencing factors, this paper proposes a short-term charging load prediction method for EV based on the GA-CNN-GRU model, which obtains the optimal combination of influencing factors' weights through the genetic algorithm (GA), and designs the data input matrix, then combines the convolutional neural network (CNN), and gated recurrent unit (GRU) to forecast the charging load of EV. The data input matrix is designed, and then the CNN-GRU model constructed by combining the advantages of CNN and GRU in feature extraction and time series prediction is used to realize accurate charging load forecasting. Ultimately, the paper confirms the efficacy of the proposed approach through practical case studies.