Simulation Research on Emergency Evacuation of Subway Station Based on Latent Class Model

In order to deeply study the influence of pedestrian psychological behavior on route choice in the emergency evacuation process of subway stations, and quantify the influence of pedestrian psychological heterogeneity on evacuation efficiency, this study establishes an emergency evacuation simulation model based on latent class model to describe the decision-making behavior of pedestrians. Firstly, the stated preference and revealed preference (SP&RP) survey method is used to investigate the psychological behavior of pedestrian groups and the behavior selection mode under different scenarios, and the latent class analysis (LCA) model is established to obtain the latent categories of pedestrians. Then, with the path length in the SP survey as the dependent variable, the distribution probability is obtained by using ordered multiple logistic regression. Finally, the evacuation environment is established in Anylogic for simulation and compared with the questionnaire results. The results show that the LCA model is in good agreement with the real observed values, and can better reflect the path choice behavior of evacuees in reality. There are differences in the influence degree of people with different psychological characteristics on evacuation efficiency. The calm type has the greatest influence on evacuation efficiency, followed by the autonomous type, and the influence of the mass type on evacuation efficiency can be ignored.