Signatures and Barcodes: Data-driven Understanding of Transportation System Performance during Extreme Events

该项目的重点是了解自然灾害等极端事件对城市交通系统的影响，这些系统是应急响应和恢复服务所必需的。由于持续的城市化和极端天气事件的频繁发生，该项目将研究在城市层面量化基础设施性能和恢复能力的新方法。项目工作成果将为负责缓解和应对极端事件的主管部门提供数据驱动的见解。将为州和地方交通官员举办研讨会，讨论这项工作的结果。该项目的部分内容将通过与伊利诺伊几何实验室的合作伙伴关系进行，以促进工程和数学专业学生的跨学科本科研究经验。该项目的重点是创建数据驱动的方法，通过解释全市和多年交通数据集来调查极端事件对交通基础设施的影响。从多元线性代数和计算拓扑的概念将被调查，以严格量化极端事件对运输系统的影响。开发的方法将用于构建“签名”，这是全市道路网络拥堵程度的可解释模式，以及“条形码”，这是网络连通性的总结。签名和条形码将被设计成更好地量化极端事件的时空影响，如偏离典型的拥堵模式和连接结构。开发的方法将应用于美国社区的大型公共交通数据集。

This project focuses on understanding the effects of extreme events such as natural disasters on urban transportation systems necessary for emergency response and recovery services. Motivated both by continued urbanization and the frequency of extreme weather events, this project will investigate novel methods to quantify infrastructure performance and resilience at city-level scales. Outcomes of the project work will provide data-driven insights relevant to authorities responsible for extreme event mitigation and response. Seminars will be given to state and local transportation officials on the results of this work. Parts of the project will be carried out via partnerships with the Illinois Geometry Laboratory to facilitate interdisciplinary undergraduate research experiences for engineering and mathematics students.This project centers on the creation of data-driven methods to investigate the effects of extreme events on transportation infrastructure by interpreting citywide and multiyear traffic datasets. Concepts from multilinear algebra and computational topology will be investigated to rigorously quantify the effects of extreme events on the transportation system. The developed methods will be used to construct "signatures," which are interpretable patterns in the congestion level of a citywide road network, and "barcodes," which summarize the network connectivity. The signatures and barcodes will be designed to better quantify the spatiotemporal effects of extreme events as deviations from typical congestion patterns and connectivity structures. The developed methods will be applied on large publically available mobility datasets in US communities.