Digital Twins for Bridge Management through the Integrating of Computer Vision and Finite Element Models

Digital Twins for Bridge Management through the Integrating of Computer Vision and Finite Element Models 3/19/2018 We propose the development of an innovative integrated solution for health monitoring of bridge structures using a non-contact image-based measurement methodology. High-resolution and regular camera systems will be used to capture the bridge movements and the traffic passing the bridge, respectively. Computer vision techniques will be used to extract the deflection time-histories of the bridge as well as the vehicle locations from the raw images. The information will be used for joint model and vehicular load identification using a Bayesian estimation technique. The updated model will be maintained as a “digital twin” of the infrastructure, which will be interrogated for health monitoring and damage diagnosis (e.g., stiffness deterioration, loss of prestress forces, etc.). The digital twin will be amenable beyond health monitoring, for global load rating, and rapid post-disaster assessment throughout the life-cycle of the bridge. Phase I of this proposal includes the preliminary implementation and verification of the computer-vision and Bayesian model updating techniques. Phase II will focus on the development of a working prototype and its validation in real-world settings. The envisioned product development stage after Phase II includes packaging the developed techniques into a practical product for near real-time monitoring and maintenance of bridges.