Logo image
Deflection estimation for structural health monitoring using computer vision algorithm
Thesis   Open access

Deflection estimation for structural health monitoring using computer vision algorithm

Ranvijay Singh
Master of Science (M.S.), Drexel University
Dec 2022
DOI:
https://doi.org/10.17918/00001452
pdf
Singh_Ranvijay_202211.85 MBDownloadView
mp4
Singh_Ranvijay_2022_Suppl15.60 MBDownloadView
Video (supplemental) Dynamic test Open Access Open Access (License Unspecified)
mp4
Singh_Ranvijay_2022_Suppl22.27 MBDownloadView
Video (supplemental) Static test Open Access Open Access (License Unspecified)

Abstract

Deflection Displacement sensor Dynamic testing Image Static test Computer Vision
Typical Structural Health Monitoring (SHM) systems are based on the use of conventional sensors designed to measure strain, displacement, acceleration, tilt, ect. Such data collection and sensing systems are robust and well known but they can be costly and require training. New opportunities are today offered by high resolution cameras that have become widely available and affordable and that are nowadays available even in smartphones. The main objective of this thesis is to explore how high resolution cameras and computer vision algorithms can be used in SHM applications, in particular, for deflection measurement of bridge structures. In the proposed structural sensing system, the collected data is obtained using a series of vision-based measures, where images collected are analyzed with computer vision algorithms to estimate structural components deflections. This vision based approach was demonstrated with 2 different controlled laboratory experiments. In one example, the deflection of a structural beam loaded by 2 hydraulic actuators was first estimated in a quasi static test. In the second example, the same approach was adopted to estimate the deflection of a steel grid (resembling a small scale bridge) subject to dynamic loading induced by a shaker. During testing, the estimated deflections were compared with measured displacements collected using traditional displacement gages and showing excellent agreement and confirming the potential for field applications on full scale structural systems. The main research significance is the validation of algorithms that can automatically analyze video images of targets placed on the structural system of interest and accurately estimate deflection of structures such as girders in bridges during load testing or during operational monitoring (deformation measured while the structure is subject to live loading from traffic).

Metrics

83 File views/ downloads
93 Record Views

Details

Logo image