Objective In recent years, hosting large-scale events has become an important means to stimulate consumption for cities and a key aspect of developing new cultural tourism. In the face of more frequent large-scale events and higher safety standards, urban rail transit operation and management are encountering greater pressure and challenges, in which passenger flow prediction during the large-scale event dispersal period is the foundation and key, necessitating focused research.

Method The "mirror image" characteristics possessed by urban rail transit passenger flow during the arrival and dissipation phases of large-scale events are analyzed. A general idea to predict urban rail transit network passenger flow during the large event dispersal period is proposed. The OD (origin-destination) demand is predicted respectively for background passenger flow and event-attending passenger flow during the event day dispersal period. A multi-path clearing model considering impedance stratification is proposed for calculating the urban rail transit network passenger flow distribution. Using the subway network of a certain city in China as a case study, a corresponding prediction software system is developed, and the predicted results for the event day dispersal period are comparatively analyzed.

Result & Conclusion The proposed prediction method achieves favorable results across three levels: network OD distribution, daily line passenger volume, and entry/exit volumes at key stations. The prediction error of the entry/exit volumes at key stations generally remains within 10% of the actual values. The established model demonstrates high accuracy in predicting whole network passenger flow diffusion and directional distribution, providing decision support for line operation plan adjustment and station passenger flow organization optimization.