Damage detection of cables in cable-stayed bridges using vibration data measured from climbing robot

Abstract

This paper presents a method for damage detection of bridge cables using the vibration signal coming from the sensor installed in the climbing robot. In this paper, the damage types are crack and local reduction in diameter. The climbing robot consisting of a body and three wheels connected to the body by three springs. The cable is considered as an axially loaded Euler beam. When there is no crack, the dynamic displacement of the robot is smooth and no distortion at the crack position can be inspected. However, when there is a damage in the cable, there is a distortion in the robot displacement at the damage location but this distortion is small and difficult to be observed visually. The distortion caused by the crack is presented as the significant peaks in the wavelet transform of the robot displacement. The proposed method is promising since it is efficient to detect cracks with depth as small as 1% of the cable diameter and to detect local changes in diameter with amplitude as small as 0.1%.

Image-based automatic multiple-defect detection of urban utility tunnel using UUTNet

“…More precisely, the network classifies each input image pixel and outputs an image containing no background and only pixels of defects. The semantic segmentation method has been widely applied in the defect detection of civil structures (Cui et al, 2021; Guo et al, 2021; Nguyen et al, 2022; Xu et al, 2022). Chen (Chen et al, 2021) proposed a two-step method, which first used the CNN model to classify cracks/no cracks in building images taken by drones, then employed the U-Net model to segment pixels of defects for images only containing cracks, thus improving the reliability and efficiency of crack detection on building facades.…”