Collaborative Research: A Deeply Integrated Physics-Based and Data-Driven Approach for Effective Resilience Management of the Power Grid

这笔拨款将开发一种新颖的、深度集成的基于物理的数据驱动方法，以评估和增强受极端气候影响的输电系统的恢复能力。美国电力基础设施越来越容易受到气候风险的影响，导致大范围和持续的停电，每年造成数十亿美元的损失。输电网自然灾害导致的故障会造成大规模和昂贵的影响，而现有的输电网扩容规划模型在很大程度上忽略了面对自然灾害时电网的弹性考虑。然而，由于数据的相对稀缺性，在配电系统弹性分析中普遍采用的纯数据驱动方法并不适用于输电系统。单方面依赖基于物理的模型也是不可行的，因为它们的计算成本极高，限制了它们扩展到网络级别的能力。这项NSF拨款旨在通过基于物理和数据驱动方法的深度整合来解决这一根本差距。本研究的结果有望帮助输电基础设施的关键决策者在各种不确定的未来情景下表征恢复能力，并确定最佳的适应或缓解策略。通过跨学科、综合研究的教育项目，对下一代学者进行灾害建模和基础设施恢复能力的教育，坚定地承诺增加学生培训的多样性，并广泛传播研究结果，作为研究项目的补充。研究方法基于大数据分析的最新发展以及基于物理的结构失效分析。以物理为导向、以数据为中心和多尺度的框架允许对气候不确定性下的网络弹性进行可扩展评估，并确定最佳投资决策。利用最先进的机器学习和计算机视觉，该项目将生成新的可公开访问的传输网络拓扑数据以及危害影响，以促进科学界在传输弹性规划方面的进一步研究。本文将建立用于输电网络失效量化的新型极限状态函数，并开发一种可扩展的结构系统不确定性量化方法。不确定过程建模的多尺度方法将有效地将传输系统数据与基础设施的计算模型融合在一起。开发的方法和数据将有助于进一步了解输电系统在未来灾害情景下的脆弱性，并能够评估投资决策对输电系统恢复力的影响。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This grant will develop a novel, deeply integrated physics-based data-driven approach to assess and enhance the resilience of power transmission systems impacted by climatic extremes. The US electricity infrastructure is increasingly prone to climatic risks that cause wide-spread and sustained outages, costing billions of dollars annually. While natural hazard-induced failures in the transmission grid lead to large-scale and costly impacts, the existing transmission expansion planning models largely neglect resilience consideration of the network facing natural hazards. The purely data-driven approaches prevalent in resilience analytics of power distribution systems, however, are not applicable to transmission systems due to relative scarcity of data. A unilateral reliance on physics-based models is not feasible either due to their extreme computational costs, limiting their ability to scale up to the network level. This NSF grant seeks to address this fundamental gap via deep integration of physics-based and data driven methods. The outcome of this research is expected to help key decision-makers for the transmission infrastructure to characterize resilience under various uncertain future scenarios and identify optimal adaptation or mitigation strategies. The research program is complemented with educating the next generation of scholars in modeling hazards and infrastructure resilience through an interdisciplinary, research-integrated educational program, a strong commitment to increased diversity in student training and broad dissemination of the results.The research approach is grounded in the latest developments in big data analytics as well as physics-based analysis of structural failures. The physics-guided, data-centric and multiscale framework allows for scalable assessment of network resilience and identification of optimal investment decisions under climate uncertainty. Using the state-of-the-art machine learning and computer vision, the project will generate new publicly accessible data on transmission network topology as well as hazards’ impacts to facilitate further research in transmission resilience planning within the scientific community. Novel limit state functions for failure quantification of transmission networks will be established and a scalable approach to uncertainty quantification of structural systems will be developed. The multiscale approach to modeling uncertain processes will effectively fuse data on transmission systems with computational models of the infrastructure. The developed methodologies and data will shed new lights on the vulnerability of the transmission system under future hazard scenarios, and enable assessing the impact of investment decisions on transmission system resilience.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.