Research on Electric Vehicle Charging Load Forecasting Method Based on Variable Mode Decomposition and one Dimensional Convolutional Neural Network

With the popularity of electric vehicles(EVs), the demand for charging infrastructure is also increasing. To enhance the precision of short-term load forecasting for EVs, a hybrid model has been developed that integrates a one-dimensional Convolutional Neural Network (1DCNN) with a Gated Recurrent Unit (GRU). This model uses channel fusion to extract time series features from different dimensional directions for model learning and fusion. At the same time, the Whale Optimization Algorithm (WOA) is used to optimize the model hyperparameters to avoid the inaccuracy of artificially set hyperparameters. This model’s performance is further enhanced by utilizing the Variable Mode Decomposition (VMD) technique to decompose the experimental data. Our proposed WOA-1DCNN-GRU model is then trained on these decomposed modal components, with the final predictions being reconstructed from them. Utilizing real-world charging pile data from a specific region, our simulation experiments confirm that the WOA-1DCNN-GRU model, which leverages VMD, demonstrates superior forecasting accuracy and stability compared to existing methods.