Pathology-based Structural Health Monitoring

forthcoming

Structural health monitoring (SHM) methods provide a comprehensive approach to the condition-based maintenance of Cultural Heritage (CH) constructions, aiming to extend their lifespan through optimized rehabilitation interventions. While physics-based damage identification is the most effective approach for comprehensive damage assessment, it has well-documented limitations. Specifically, classical methods often rely on a single model of the CH asset, which introduces significant uncertainties in decision-making. To address this challenge, this Action introduces a pioneering SHM paradigm referred to as pathology-based SHM (Path-B SHM) using multi-class digital twins (DTs)—a population of competing, efficient meta-models. The methodology begins with a systematic diagnosis, including site visits supplemented by non-destructive testing (NDT), to identify the most probable structural pathologies the asset may experience throughout its lifespan. High-fidelity scenario-based numerical models are then developed for each identified pathology. To enable real-time damage identification and simultaneous inverse calibration of all scenario-based models, these numerical models are replaced by computationally efficient meta-models constructed using artificial intelligence (AI) tools which form a multi-class DT. The resulting DT is capable of inferring, in quasi real-time, the probability of occurrence of the identified potential damage pathologies and their severity when supplied with continuous monitoring data. If an anomaly is detected, a Bayesian model selection approach identifies the damage mechanism, selecting the meta-model that explains the experimental data with the highest likelihood. This approach, therefore, enables rapid, robust, and comprehensive damage identification. Path-B SHM has the potential to revolutionize and advance the frontiers of SHM, offering unprecedented opportunities for data-informed decision-making in the preservation of CH constructions.