Learning Failure Propagation Patterns in Interdependent Network From Observed Post-Disaster Disruptions

该项目旨在开发模型和工具，以揭示故障级联过程，或者初始的小节点集的故障如何通过相互依赖的网络传播，就像在现实世界中发生的那样。最近的灾难，如美国和加拿大的停电，以及2012年的飓风桑迪，一再表明了相互依存的关键基础设施之间的级联故障对社会的影响。本研究解决了构建下一代弹性关键基础设施的基本挑战：系统内部和系统之间的哪种相互依赖关系将促进或减轻潜在的故障级联过程。这个项目也将提供一个重要的机会来开发一个结合多个领域知识的方法论课程。课程中开发的模块将通过项目网站广泛传播。外联活动将重点关注代表性不足的少数民族学生，以及一个由灾害/紧急反应从业人员组成的咨询小组。该项目将利用新颖的地质统计学和统计学方法，使pi能够在精细的空间尺度上估计灾害破坏，并使用灾难前后收集的经验数据确定复制故障传播过程的事件序列。它还将使用独立的数据集、报告和非正式的定性访谈来统计验证和背景化数值模型结果。该项目建立在坚实的知识和方法基础之上，这些基础来自于我们以前的研究，以及高级统计、一般网络科学和基础设施相互依赖性方面的最新发展。这项研究有可能在多个领域带来革命性的变化。

This project is to develop models and tools to uncover the failure cascading process, or how failures of an initial, small set of nodes propagate through interdependent networks as it happened in the real world. Recent disasters such as blackouts in the United States and Canada and the 2012 Hurricane Sandy have repeatedly shown the effects on society resulting from cascading failures between interdependent critical infrastructures. This research addresses a fundamental challenge in building toward our next-generation resilient critical infrastructures: what kinds of interdependencies within and across systems will facilitate or mitigate a potential failure cascading process. This project will also provide a critical opportunity to develop a methodological class that combines knowledge from multiple fields. The modules developed in the course will be widely disseminated through the project website. Outreach activities will focus on under-represented minority students, as well as an advisory team comprising practitioners in disaster/emergency response. This project will utilize novel geostatistical and statistical approaches that will allow the PIs to estimate disaster disruptions at a fine spatial scale and identify a sequence of events replicating the failure propagation process using empirical data collected before and after a disaster. It will also use independent datasets, reports and informal qualitative interviews to statistically validate and contextualize the numerical model results. This project builds on a solid intellectual and methodological foundation resulting from our previous studies as well as the most recent developments in advanced statistics, general network science, and infrastructure interdependency. The research has the potential to bring transformative changes across multiple fields.