Operational Health Monitoring of Bridges Using Bayesian Model Updating and Computer Vision Techniques

Abstract

This study presents a new framework for operational health monitoring of bridges using computer vision and finite element (FE) model updating techniques. In this framework, the bridge acceleration responses and vehicle tracking data are obtained under operational conditions using accelerometers, regular traffic cameras and computer vision techniques. These data are synchronized and integrated with the initial FE model of the bridge through a FE model updating process to jointly estimate the dynamic load of tracked vehicles on the bridge and the unknown FE model parameters. The final estimates of FE model parameters reveal information regarding the location and extent of potential damage in the monitored bridge. This framework is first successfully verified in a numerically simulated environment and then validated using a real-world bridge subjected to traffic excitation due to passage of tri-axle dump trucks.