Bayesian-Based Fusion of Monitoring Data and Visual Inspections in Monumental Structures

The growing need for assessing the integrity of aging monumental structures by means of cost-effective and not-destructive techniques has driven significant scientific interest in vibration-based structural health monitoring (SHM) procedures, allowing to track changes in selected damage-sensitive structural parameters. However, the evaluation of a healthy or damaged state from the acquired monitoring data is often a data-driven process that can be subjected to a large amount of uncertainty, resulting in false positives and false negatives. Hence, the main idea behind this work focuses on handling the main issues related to the uncertainty management by exploiting the aggregation of different sources of information. In this context, this study is aimed at detecting and locating structural damages in monumental structures with the aid of a data fusion approach including vibration-based system identification, Bayesian-based finite element (FE) model updating and visual inspections. As a preliminary step, potential damage-sensitive sections are defined on the basis of nonlinear static analyses (NLSA) performed on a calibrated FE model and/or engineering judgment (EJ). Then, a surrogate model is established enabling to transfer knowledge from the monitoring data to the updated numerical models and to solve the inverse problem aimed at deriving the posterior statistics of the uncertain parameters. The effectiveness of the proposed approach is demonstrated by using 1-year of recorded data acquired in a monumental structure named Consoli Palace, located in Umbria, central Italy, a region characterized by high seismic hazard. The palace has been continuously monitored by the Authors since 2017 using dynamic, static and environmental sensors.