Vulnerability Assessment of Bridge Networks Integrating Model Updating and Bayesian Network Modeling

Structural health monitoring (SHM) is a powerful tool for post-earthquake seismic vulnerability analysis, contributing significantly to infrastructure resilience. However, equipping all assets and structural elements with on-site instrumentation is often neither economically feasible nor practical, and typically only a limited number of structures within the network are monitored. Therefore, integrating machine learning techniques with the existing SHM systems to transfer and extrapolate information across monitored and non-monitored assets is an attractive solution. In this paper, we propose a Bayesian Network (BN) as an ideal tool for inference under conditions of limited data, combined with a structural model updating technique. This approach enables extending inferences from monitored assets to similar, unmonitored structures within a network. To demonstrate the effectiveness of the proposed framework, a synthetic bridge network modeled by a set of finite element models and subjected to an earthquake scenario is considered. Using the framework, fragility curves for all the assets in the network are updated based solely on the data received from a set of monitored assets. This illustrates the potential of the proposed method in improving the accuracy of vulnerability assessments while reducing the need for extensive sensor deployment.