Bayesian Inference and Gaussian Process Regression for Structural Health Monitoring of Box Girder Bridges Joints

Abstract

Structural Health Monitoring (SHM) plays a crucial role in ensuring the safety and availability of infrastructure, especially for bridge structures. The existing methods often fail to detect early-stage or localized damage, which necessitates the development of methods that can provide accurate and real-time monitoring. This study introduces a Bayesian Inference-based framework enhanced by Gaussian Process Regression (GPR) for assessing the structural health of segmental box girder bridge joints. Finite Element Model (FEM) updating is used in conjunction with Bayesian methods. The Latin Hypercube Sampling (LHS) technique is used to establish a comprehensive dataset of elastic modulus values, later analyzed with the Markov Chain Monte Carlo (MCMC) approach. The results show that the method is successful in determining the localized damage for precast segmental box girder bridges, with posterior distributions closely matching actual elastic modulus values. The proposed Bayesian model updating technique shows significant potential in enhancing the detection of structural damage with reduced computational cost, which provides a practical and efficient tool for real-time monitoring of structural health in box girder bridges.