Prediction of EV charging load based on federated learning

With the rapid development of electric vehicles (EV), the sharp increase in the number of charges and the daily charging volume has a impact on the stable operation of the power grid. Therefore, the study of EV charging load forecasting is of great significance. However, due to the privacy of charging behavior data, existing research based on machine learning prediction models fails to consider this important factor, resulting in low prediction accuracy.To solve this problem, this paper proposes a power prediction model based on federated learning and variational mode decomposition-long short-term memory neural network (VMD-LSTM). Firstly, VMD is to decompose the EV charging power time series into multiple components for hierarchical prediction, which reduces the non-stationarity and complexity of the EV charging sequence. Secondly, an improved particle swarm optimization (PSO) algorithm is used to improve the decomposition efficiency of VMD, and horizontal federated learning is realized through local training and parameter aggregation methods, data privacy security while predicting the EV charging load. Finally, the proposed method is verified using charging load data from multiple charging stations in a city. The results that the proposed method effectively improves the accuracy of short-term EV charging load prediction while ensuring user privacy security.