Completely non-contact modal testing of full-scale bridge in challenging conditions using vision sensing systems

Abstract

Vision-based modal testing method can provide a contactless approach to understanding dynamic properties of bridges. However, the applicability of this method to in-service bridge measurement has been limited by its drawbacks, such as low-accuracy measurement data of entire bridges, low robustness to varying lighting, artificial target installation or distinct natural features requirement, etc. In addition, the existing vision-based methods rarely captured torsional modes of the real-life bridges, leading to an insufficient understanding of the fundamental dynamic properties of the structures. This paper proposes a new vision-based modal testing framework to overcome these drawbacks. A multiple-setup strategy with three time-synchronised cameras is used to record videos of dynamic structural motion. A robust object tracking technique is used to extract reliable displacement of the field structure from the videos with significant noises. This framework provides a completely non-contact approach to capture bending and torsional modal properties of full-scale bridges in uncontrolled field conditions. A large-scale laboratory bridge and in-service suspension footbridge were selected to verify the feasibility of the proposed approach to extract correlated modal properties. The modal analysis results showed a satisfactory agreement with the accelerometer measurement, demonstrating the proposed framework's efficacy in operational modal analysis.