Emergency evacuation behavior RPL and simulation research in public transport hub with the consideration of passenger heterogeneity

Heterogeneous passenger behaviors greatly influence the safety and order of emergency evacuation, which merits in‑depth investigation. Based on questionnaire data, this study analyzes passengers’ demographic, psychological and cognitive attributes, where psychological and cognitive characteristics are defined as latent variables. Structural equation modeling (SEM) is adopted to estimate these latent variables, with a weighted path diagram presenting indicator factor loadings and latent‑variable path coefficients. Incorporating latent variables, exit distance and passenger flow‑direction are set as random parameters, and qualified indicators are selected as interaction terms to develop a random‑parameters Logit model accounting for passenger heterogeneity. Taking Beijing Songjiazhuang Subway Station as the case, AnyLogic simulates emergency evacuation scenarios. Comparisons of density heatmaps and evacuation times against the baseline model verify that the proposed model better reproduces real‑world evacuation processes and yields higher evacuation efficiency. Relevant suggestions are finally provided to guarantee passenger safety and evacuation performance in public transport hubs.