Evaluation of the resilience of large-scale interdependent infrastructure through simplification-enhanced network modeling
提出一个基于网络的框架，通过针对不同基础设施类型的简化策略（如两阶段重聚类和特征骨架化），在保持高精度的同时大幅降低计算时间，并用于评估气-电-交通系统在多灾害场景下的韧性。
With the growing impacts of extreme disasters, interdependencies among urban critical infrastructure systems (CISs) introduce additional vulnerabilities to system resilience. While network-based approaches effectively model intrinsic interdependencies, the computational complexity of large-scale systems necessitates effective structural simplification. To address this, this paper introduces a comprehensive network-based framework that explicitly integrates interdependency modeling into a systematic resilience assessment workflow. A key enabler of this framework is tailored network simplification strategies for different infrastructure types, including a two-stage reclustering method to mitigate the impact of topological errors and a feature-informed skeletonization strategy for road networks. In the validation case study using a real-world gas-electricity-transportation system, these strategies achieved approximately 80% and 18% reductions in CPU time, respectively, relative to the detailed networks while maintaining high accuracy in resilience estimation. The validated framework was then applied to assess the system's resilience under a detailed multi-hazard scenario. The results reveal that interdependencies degrade individual subsystem resilience, make overall recovery dominated by the slowest-recovering subsystem, and significantly alter the spatial distribution of vulnerabilities. These findings highlight the necessity of explicit interdependency modeling, and the proposed method offers a practical and integrated solution for the resilience assessment of large-scale, interdependent CISs.