CAREER: A Hybrid Physics-Based/Data-Driven Modeling and Mitigation Approach for Interdependencies Between the Electric Power, ICT, and Transportation Critical Infrastructures

提高关键基础设施 (CI) 的保护能力和抵御自然灾害和人为威胁的能力是一项迫切的短期目标。三个高度相互依赖的 CI 是本研究的重点：电力系统、信息和通信技术 (ICT) 以及电动车辆/交通（电子交通）。管理人员需要对故障模式和传播进行网络建模，以制定策略来减轻相互依赖的 CI 之间的故障影响。尽管迄今为止在 CI 的网络建模方面已经取得了很大进展，但这些努力未能捕捉到 CI 的复杂性，基于不充分或不准确的假设，忽略了相互依赖性，并且大多是经验和统计驱动的，而不是基于物理原理。本研究的目的是改进相互依赖的 CI 的故障和弹性建模。然后，这项工作将利用强化学习来帮助决策进行扩展。结果将是一个框架，用于解释 CI 相互依赖性，这对于故障和灾难规划和恢复至关重要。智力价值。基于影响图的模型和基于强化学习的相互依赖性缓解框架雄辩地解决了长期挑战城市/基础设施规划者和运营商的一个关键问题：如何建模、预测和防止通过多个关键基础设施传播的级联故障。当前的方法缺乏准确性和处理复杂性和相互依赖性的能力。这项工作应用了详细的基于物理的自下而上的建模以及强化学习的创新应用，为该领域提供了变革性的进步，从而为 CI 设计、管理、灾难规划和恢复提供了更加灵活和准确的分析方法。 更广泛的影响。该项目将通过降低基础设施的运营成本、减少停电的可能性和影响以及改善城市对重大骚乱的反应来对社会产生积极影响。这些教育部分将使来自纽约市立大学和邻近高中的数百名学生受益，这些高中招收了大量在 STEM 领域代表性不足的学生。这将增加具有多学科背景的训练有素的工程师和科学家的数量，他们将在定义社会的未来方面发挥重要作用。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

Improving the protection and resilience of critical infrastructures (CIs) against natural disasters and manmade threats is an imperative short-term goal. Three highly interdependent CIs are the focus of this research: electric power systems, information and communication technologies (ICT), and electricity- based vehicles/transportation (e-transportation). Network modeling of failure modes and propagation is needed by managers to devise strategies that mitigate the impact of failures across interdependent CIs. Although much progress has been made in network modeling of CIs to date, these efforts fail to capture the complexity of CIs, are based on inadequate or inaccurate assumptions, neglect interdependencies, and are mostly empirical and statistically driven rather than based on principles of physics. The objective of this research is to improve failure and resilience modeling of interdependent CIs. This work will then be expanded using reinforcement learning to aid in decision-making. The outcome will be a framework to account for CI interdependencies that are critical for failure and disaster planning and recovery.Intellectual Merit. The influence graph-based model and reinforcement learning-based interdependency mitigation framework eloquently address a key problem that has long challenged city/infrastructure planners and operators: how to model, forecast, and prevent cascaded failures that propagate through multiple critical infrastructures. Current approaches lack accuracy and the ability to handle complexities and interdependencies. This work applies detailed physics-based bottom-up modeling coupled with the innovative application of reinforcement learning to provide a transformative advancement to the field, resulting in a much more flexible and accurate analysis methodology for use in CI design, management, disaster planning, and recovery. Broader Impact. This project will positively impact society by reducing the operational cost of infrastructures, decreasing the likelihood and impact of blackouts, and improving cities' responses to major disturbances. The educational components will benefit hundreds of students from CUNY and neighboring high schools that enroll significant numbers of students who are underrepresented in STEM fields. This will increase the number of highly trained engineers and scientists with multidisciplinary backgrounds who will have a major role in defining the future of society.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.