Monocular Vision-Based Key Dynamic Parameters Measurement Method Used for Typical Bridge Structures Health Monitoring

Abstract

The key dynamic parameters measurement of bridge structures is essential to their health monitoring and has been highly valued. However, the commonly used accelerometer- or linear variable displacement transducer (LVDT)-based measurement methods have inevitably appeared in issues such as: measurement inconvenience, poor real-time performance, high cost, and low accuracy. In this study, a new monocular vision (MV)-based key dynamic parameters measurement method used for typical bridge structures health monitoring is proposed, which is capable of improving the performance with a commercial camera. This method combines the reliable camera calibration with a simple ring mark and improved sub-pixel Zernike moment edge extraction with linear ramp gray-scale model to accurately measure the key dynamic parameters. The laboratorial comparison experiments with the current accelerometer-based method (AM) at different working conditions to measure the acceleration, frequency, and deflection at the mid-span of a self-built bridge model were performed. The maximum relative deviations of the acceleration amplitude, frequency and deflection measured in the mid-span between the accelerometer-based and proposed methods were 0.52%, 1.35%, and 2.51%, respectively. Additionally, the comparison experiments on a practical bridge were also accomplished, and the results demonstrated that the proposed method can obtain a considerable measurement performance.