Adaptive structural–operational resilience intelligence for BIM-enabled infrastructure systems: a causal graph neural and explainable AI framework for hidden vulnerability detection and disaster-resilient governance
TL;DR
Methodologically, it advances BIM-enabled resilience intelligence by integrating causal reasoning, multiplex graph learning, Bayesian uncertainty quantification and XAI within a unified framework for proactive disaster-resilient infrastructure governance and decision support.
Abstract
This study aims to propose an adaptive structural–operational resilience intelligence architecture for building information modeling (BIM)-enabled infrastructure systems to model resilience dissipation, hidden vulnerability accumulation and adaptive instability propagation across interacting infrastructure subsystems. This study addresses limitations of conventional BIM-enabled infrastructure analytics that primarily emphasize isolated degradation prediction, static resilience assessment and limited support for adaptive resilience governance. The proposed framework integrates BIM, structural causal modeling, multiplex infrastructure interaction networks, graph neural networks, Bayesian uncertainty analytics and explainable artificial intelligence (XAI) within a unified resilience intelligence architecture. A publicly available BIM–artificial-intelligence-integrated infrastructure data set containing structural, operational, environmental, governance, anomaly intelligence, resource and risk-related variables was used to evaluate subsystem-interactions, resilience deterioration dynamics and governance-sensitive infrastructure behavior. The results demonstrate that infrastructure instability evolves through synchronized interactions among structural degradation, operational stress, environmental exposure and governance instability rather than isolated structural failure alone. The proposed structural–operational criticality (SOCI) mechanism successfully identified adaptive instability amplification and resilience dissipation across dynamic coupling regimes. XAI analysis identified Safety_Risk_Score as the dominant resilience deterioration driver, while counterfactual governance analysis demonstrated measurable resilience improvements through adaptive operational stabilization and governance reinforcement interventions. This study contributes theoretically by introducing co-evolutionary infrastructure theory, resilience dissipation dynamics and SOCI to explain adaptive infrastructure instability evolution. Methodologically, it advances BIM-enabled resilience intelligence by integrating causal reasoning, multiplex graph learning, Bayesian uncertainty quantification and XAI within a unified framework for proactive disaster-resilient infrastructure governance and decision support.
