A portable monitoring approach using cameras and computer vision for bridge load rating in smart cities

Abstract

Smart structures require novel, efficient, and effective technologies for their safe operation and serviceability. This paper presents a novel, practical, cost-effective, and field test-based methodology using portable cameras and computer vision technologies to identify the lateral live load distribution factors for the existing highway bridges to perform load rating. By using a computer vision-based measurement method and traffic recognition, the girder deflection under live load can be monitored in a noncontact way and can be utilized to derive the load distribution. To verify the feasibility of the proposed approach, a comparative experimental study is conducted on a real-life pre-stressed concrete bridge with a set of conventional load tests and experiments in normal traffic. The results are compared with the conventional approach, such as simplified formulations recommended by AASHTO specifications, and the experimental method using the data from strain gauges and a calibrated finite element model (FEM). The comparative results show that the proposed approach can obtain very similar load distribution factors and bridge load rating factors both in a conventional load test and normal traffic. In comparison to the simplified formulation recommended by AASHTO specifications, the proposed approach can reflect the real-life structural properties and improve the load rating factor of AASHTO specifications by around 12%. In addition, as compared to the load-test-based approaches, such as using strain data and calibrated FEM, the proposed approach does not require traffic closure and a large amount of effort to deal with the load test and model updating. The bridge studied in this paper represents a very typical one from a large population of bridges that are part of the smart infrastructure. Such a practical approach will be practical and cost-effective for bridge load rating in smart cities.