Harnessing Network Dynamics for "Smart" Infrastructure and Urban Resilience

动机：大多数大型复杂系统在发生连锁故障时都存在固有的脆弱性。在这些过程中，最初的小干扰被允许传播和放大，最终导致系统的大部分改变行为或发生故障。例子包括停电、物种灭绝浪潮和金融恐慌。因此，梯级预报和防御在一系列应用中至关重要。研究问题：几十年的研究已经提供了大量关于连锁故障的风格化见解，包括生成机制和统计特性。然而，许多实际应用需要预测单个事件的能力，这反过来又需要一些关于系统初始状态和详细动态的知识。后者是研究“系统系统”(如电力+运输系统)中的级联的特殊障碍，对于不同系统之间的耦合的详细建模理解仍然难以捉摸。研究目标：本研究计划的目标是加深对连锁故障的动力学基础的理解，特别是在多层系统中。我们将通过三个相互关联的短期研究目标来实现这一点，使用基础设施网络作为激励应用：1)了解电力/运输系统中的个人电力/运输故障如何导致人类行为的变化；2)量化可观察性的实际极限-我们(或对手)确定的能力；最后3)设计基于时变网络的自适应、自我修复的电力系统。预期结果：到拟议的5年研究计划结束时，我们将确定可用于预测、监控和抵消基础设施系统内和基础设施系统之间的大规模故障的特定动态功能。我们的长期目标是建立在这些见解的基础上，朝着对城市基础设施进行准确的“系统系统”建模的方向发展，该模型考虑了不同基础设施层之间的结构和动态异质性以及人类行为的影响。影响：这里产生的见解将推进复杂系统中动态推理、故障建模和控制的技术水平。在此过程中，它将为工程师和政策制定者提供新的工具，以加强城市基础设施，以抵御年龄、气候变化和蓄意攻击带来的威胁。此外，这些工具预计将提供对相关网络问题的见解，包括可持续的生态系统和金融系统管理。

Motivation: Most large complex systems possess an inherent vulnerability to cascading failures. In these processes, an initially small disturbance is allowed to propagate and amplify, eventually causing large parts of the system to change behavior or fail. Examples include power outages, waves of species extinctions, and financial panics. As such, cascade forecasting and defense is of paramount importance in a range of applications. Research Problem: Decades of research have provided a trove of stylized insights about cascading failures, including generative mechanisms and statistical properties. Yet many practical applications demand the ability to forecast individual events, which in turn demands some knowledge the systems initial state and detailed dynamics. The latter represents a special obstacle to the study of cascades in "systems of systems" (such as the power + transportation systems), where detailed modeling understanding of the couplings between different systems remains elusive. Research Objectives: The goal of this research program is to generate deeper understanding of the dynamical foundations of cascading failures, specifically in multilayer systems. We will pursue this through three interrelated short-term research objectives, using infrastructure networks as a motivating application: 1) understanding how individual power/transportation failures in power/transportation systems induce changes in human behavior; 2) quantifying the practical limits of observability-our (or an adversary's) ability to determine ; and finally 3) designing adaptive, self-healing power systems based on time-varying networks. Anticipated Outcomes: By the end of the proposed 5-year research program, we will have identified specific dynamical features that can be exploited to predict, monitor, and counteract large-scale failures both within and between infrastructure systems. Our long-term goals are to build on these insights, moving toward accurate "systems-of-systems" modeling of urban infrastructure that accounts for the structural and dynamical heterogeneities between different infrastructure layers as well as the influence of human behavior. Impact: The insights generated here will advance the state of the art of dynamical inference, failure modeling, and control in complex systems. In the process, it will provide engineers and policymakers with new tools to fortify urban infrastructure against the threats posed by age, climate change, and deliberate attack. In addition, these tools are expected to provide insights into related network problems, including sustainable ecosystem and financial system management.