Arrival Information-Guided Spatiotemporal Prediction of Transportation Hub Passenger Distribution

Abstract

Understanding the spatiotemporal distribution of hub passenger flow is essential for both hub and urban transportation operations. However, predicting spatiotemporal distribution of transportation hub passenger flow encounters challenges due to intricate factors influencing its dynamics. We propose a model, the Spatiotemporal Multi-Graph Convolutional Network (STMGCN), to predict the spatiotemporal distribution of hub passenger flow in urban areas. The model comprises a spatiotemporal passenger flow prediction module and a passenger flow correction module. The spatiotemporal passenger flow prediction module integrates a Graph Convolutional Network (GCN) and a Gated Recurrent Unit (GRU). These components consider the influences of region functionality, adjacency, distance to the hub, and weather. The passenger flow correction module determines reserved passenger flow based on future and historical hub arrival passenger similarities. The results from both modules are combined to obtain the predicted passenger flow. In a case study using Beijing Daxing International Hub in China, the results demonstrate that the STMGCN model outperforms baseline models in prediction accuracy. Furthermore, the model exhibits stability in predictive performance, particularly in regions with significant changes in passenger flow and during holidays. Meanwhile, we demonstrated that traffic analysis zone functionality and arrival passenger flow play significant roles in passenger flow prediction.

Keywords: Hub passenger flow, Spatiotemporal distribution, Deep Learning, Multi-graph based model, Arrival information