Load forecasting of electric vehicle charging stations is crucial for the stable and efficient operation of the power system. High-precision charging station load forecasting can optimize the operation and planning of the power system to provide continuous service. Although existing load forecasting methods have made some progress, they still face challenges such as data sparsity and spatial heterogeneity. To address these two challenges, we propose a Parallel Spatio-Temporal Graph Network Model (PSTGNM). The model combines the Temporal Graph Convolutional Network (T-GCN) and the Spatio-Temporal Fusion Graph Neural Network (STFGNN). Specifically, T-GCN effectively captures the spatiotemporal dependencies in graph-structured data, while STFGNN can capture more complex spatiotemporal dependencies. We then further improve the overall performance of the model by fusing the output information of the two networks using a gating mechanism. Experimental results on four real-world electric vehicle charging station datasets show that PSTGNM outperforms other baseline methods, which demonstrates the effectiveness of PSTGNM in handling data sparsity and spatial heterogeneity.
Tian et al. (Mon,) studied this question.