A Human–AI Integrated Framework for Resilience Modeling of Urban Infrastructure under Extreme Conditions

Urban infrastructure systems, such as subway networks and water supply systems, are vulnerable to disruptions from extreme events. This study addresses the challenge of representing how humans detect and resolve abnormal states, enabling computers to synthesize operational experiences under incomplete information. A modeling framework is proposed to formalize (1) physical facility elements; (2) classified abnormal states; and (3) human observation, analysis, and action activities for anomaly resolution. Using schemas, the framework incorporates foundational components—tanks, pipelines, controllers, and sensors—characterized by attributes like pressure, flow, and storage. It captures critical relationships and behaviors using ontological modeling principles. The framework is applied to subway flooding simulating varying inflow rates, structural damage, and multiple inlets. Results demonstrate its ability to represent abnormal states, assess resilience strategies, and support human–AI collaboration. This work provides a foundation for modeling infrastructure resilience and its limitations.