Multi-class traffic mass evacuation optimization considering non-compliance behavior
针对疏散中人们不服从政府指定路线的问题，建立数学模型并设计算法，在2024年能登半岛地震案例中使总疏散时间减少6%以上，为制定更现实的疏散策略提供依据。
Effective evacuation management is essential for minimizing casualties and economic losses, as natural disasters have devastating impacts on lives, infrastructure, and economies worldwide. However, traditional evacuation models often assume full compliance with government-issued evacuation orders, overlooking the reality that a significant proportion of evacuees may not follow these directives. This study addresses this challenge by developing a comprehensive mathematical model that investigates the impact of evacuees’ non-compliance behavior with government-assigned evacuation routes on overall system evacuation efficiency. To support this, a systematic survey is conducted under hypothetical disaster scenarios to investigate the factors influencing evacuees’ non-compliance behavior. The key influencing factors identified through the survey are then incorporated into a logit-based multi-class traffic assignment model. To address the non-convex nature of the model, a customized adaptive variable neighborhood search algorithm is developed. We evaluate the proposed multi-class evacuation optimization framework using case studies of two cities affected by the Noto Peninsula earthquake on January 1, 2024: Wajima City and Shika Town. Compared with the scenario that ignores non-compliance behavior, the proposed model reduces total evacuation time (TET) by 7.14% in Wajima City and 6.36% in Shika Town. Average evacuation time (AET) and longest evacuation time (LET) also decrease consistently (Wajima City: AET -7.17%, LET -3.53%; Shika Town: AET -6.37%, LET -6.30%), demonstrating the benefits of incorporating realistic behavioral assumptions into evacuation planning. This study provides valuable insights for policymakers and disaster planners, enabling the design of more resilient and adaptive evacuation strategies that account for human behavioral variability during emergencies.