Optimization of passenger evacuation path in flood scenarios considering companion behaviors

Abstract

The optimization of emergency evacuation paths for passengers in underground rail transit hubs has become a critical challenge in urban flood prevention and disaster mitigation systems. Most previous evacuation path optimization methods assume passengers move independently as individuals without considering socially connected groups traveling together. To address this, this paper proposes a novel passenger evacuation path optimization method considering companion behavior during subway station flooding incidents, and develops an innovative ETACO algorithm to solve the path optimization model. Taking an actual subway station as a case study, a station simulation system constructed using PathFinder is employed to simulate passenger evacuation processes, demonstrating the effectiveness of the proposed path optimization scheme. An improved entropy weight method is introduced to conduct a multidimensional evaluation of evacuation performance. The results indicate that: (1) Companion behavior significantly inhibits evacuation efficiency, with higher proportions of grouped evacuees leading to increased evacuation time and reduced average movement speed; (2) The proposed ETACO dynamic optimization strategy remarkably enhances the convergence performance of the path optimization model solution, achieving a 16% improvement in average optimal objective improvement rate compared to conventional ACO, while generating more efficient path optimization strategies; (3) Increasing numbers of interrupted road sections progressively slow down passenger evacuation; (4) Evacuation effectiveness evaluation further verifies the enhancement of station safety performance through the path optimization strategy. This research provides a more realistic solution for evacuation path optimization considering companion behavior in complex flood scenarios.