Analyzing evacuation behaviors using virtual simulation and Levenshtein similarity: A case study of railway stations

Effective analysis of evacuation behavior is essential for improving safety management in complex public spaces such as railway stations. This study presents a methodological framework that integrates a virtual simulation environment with the Levenshtein Similarity method to quantitatively examine behavioral sequences during emergency evacuation scenarios. A controlled experiment was conducted using a repeated measures design to observe participants' route choices under both normal and simulated emergency conditions. Behavioral trajectories were compared using Levenshtein Similarity to identify patterns and deviations in decision-making processes. The results demonstrate that this combined approach captures variations in individual responses to environmental cues, such as exit signs and spatial configurations. By focusing on low-congestion scenarios, the method provides a robust and reproducible way to assess evacuation strategies and decision-making processes. The study highlights the potential of integrating simulation and similarity based analysis as a scalable tool for evaluating human behavior in safety-critical environments.