Charging-Load Prediction for Electric Vehicle Stations Using Correlation Analysis and RIME-CNN-LSTM-Attention Model

The rapid growth of electric vehicles (EVs) has made accurate forecasting of charging station loads essential for ensuring grid stability and supporting infrastructure planning. While previous studies have investigated this problem, correlation-based feature selection remains relatively underexplored, which may lead to redundant features and reduced prediction accuracy. To address this issue, this paper proposes a hybrid forecasting model, RIME-CNN-LSTM-Attention, which integrates correlation-driven feature selection with advanced deep learning. Pearson, Spearman, and Kendall's tau-b analyses are first applied to identify the most influential factors affecting charging demand. The RIME algorithm is then used to optimise the hyperparameters of the CNN-LSTM network, while the attention mechanism dynamically emphasises critical load fluctuation periods. Case studies utilising actual charging station data illustrate that the proposed model substantially surpasses benchmark methodologies, thereby improving the accuracy and resilience of electric vehicle charging load forecasting.