Employing vision-based sensing for long-term monitoring
Kassotakis, N; Burn, N; Fenney, L; et al.Pillai, A; Johanning, L; Kassotakis, N
Date: 16 August 2022
Conference paper
Publisher
NDT-CE
Related links
Abstract
Despite the crucial role of structural health monitoring (SHM), traditional approaches rely on contact-based sensors which are both costly and lack automation. Vision-based sensing techniques such as Digital Image Correlation (DIC) have recently emerged as a viable substitute, due to their non-contact nature and low cost. To date, ...
Despite the crucial role of structural health monitoring (SHM), traditional approaches rely on contact-based sensors which are both costly and lack automation. Vision-based sensing techniques such as Digital Image Correlation (DIC) have recently emerged as a viable substitute, due to their non-contact nature and low cost. To date, however, the long-term performance of DIC has not been evaluated. This study assesses DIC for long-term displacement monitoring. Firstly, the robustness of the monitoring of ambiently excited structures over long periods is examined. This is achieved through the measurement of drift of control points. Then, correlation is examined between the drift measurements and the ambient temperature, to examine the influence of temperature on the robustness of the DIC measurements. After, the effectiveness of employing DIC for monitoring ambiently excited structures is examined. Towards this aim, the displacements of the midspan of the experimental bridge structure are monitored for one month and compared with those of a traditional contact-based sensor, i.e., a Linear Variance Displacement Transducer (LVDT). Finally, to further demonstrate the effectiveness of employing DIC measurements for ambiently excited structures, a correlation is sought between the midspan displacements and the ambient temperature. Concerning the robustness of the long-term DIC measurements, the drift was found to be relatively small (i.e., equal to 0.06 mm) whilst the temperature was found to potentially influence this. With regard to the overall effectiveness of long-term monitoring with DIC, the study found that non-contact sensing has comparable accuracy to the LVDT, with a correlation coefficient equal to 0.996, root mean square error of 0.012 mm, and mean absolute error of 0.010 mm. Moreover, the correlation of DIC measurements with temperature showed its effectiveness in capturing complex structural behaviours (e.g., extremely slow and small movements) typically associated with ambiently excited structures. Whilst this study is only on a small-scale structure, it paves the way for the employment of vision-based on large-scale structures enabling the general use of DIC for SHM.
Engineering
Faculty of Environment, Science and Economy
Item views 0
Full item downloads 0