During disasters, swiftly and efficiently evacuating populations in hazardous situations is crucial to minimize losses. This study proposes a novel framework to address dynamic population evacuation (DPE) problems, which includes planning and online evacuation management phases facilitated by vehicular communication. In the planning phase, a shelter allocation problem (SAP) is solved dynamically for destination choice and a dynamic traffic assignment (DTA) for path choice toward the chosen destinations to obtain an initial evacuation plan. The initial plan is then enhanced by employing a vehicular ad hoc network (VANET) within the vehicular edge computing (VEC) architecture. This enhancement enables communication among evacuees, allowing them to revise their vehicle's route choice and planned destination. These revisions take into account the changing risk and traffic conditions. The proposed online DPE framework is applied to the real evacuation scenario of Mill Valley City, CA. The proposed model is evaluated with different VANET architectures, including vehicular cloud computing (VCC) and VEC. The results show that the VEC framework outperforms other configurations and improves the evacuation process compared to the scenario with an initial plan by more than 30% in network clearance time. Additionally, a performance analysis is carried out for evacuation scenarios with different penetration rates of connected vehicles in VANET.

中文翻译：

---

利用车载通信系统进行智能动态疏散规划和在线管理

在灾害期间，迅速有效地疏散危险情况下的人员对于最大限度地减少损失至关重要。这项研究提出了一个解决动态人口疏散（DPE）问题的新颖框架，其中包括由车辆通信促进的规划和在线疏散管理阶段。在规划阶段，动态解决避难所分配问题（SAP）以进行目的地选择，并动态解决动态交通分配问题（DTA）以选择通往所选目的地的路径，以获得初始疏散计划。然后通过在车辆边缘计算 (VEC) 架构中采用车辆自组织网络 (VANET) 来增强初始计划。这一增强功能使疏散人员之间能够进行通信，使他们能够修改车辆的路线选择和计划目的地。这些修订考虑了不断变化的风险和交通状况。所提出的在线 DPE 框架应用于加利福尼亚州米尔谷市的真实疏散场景。所提出的模型使用不同的 VANET 架构进行评估，包括车辆云计算 (VCC) 和 VEC。结果表明，VEC 框架的性能优于其他配置，与初始计划的场景相比，疏散过程在网络清理时间上提高了 30% 以上。此外，还针对VANET中不同联网车辆渗透率的疏散场景进行了性能分析。