Session: 05-03: Vibration Based Methods

Paper Number: 88421

88421 - Vibration-Based Bridge Damage Detection Using Image-Based Pre-Trained Deep Learning Network 

Vibration-based damage detection has become one of the principal practices to prevent the structural collapses in civil, mechanical, and other engineering disciplinaries. Meanwhile, with the advancement of computing technology, various machine learning (ML) approaches have been paid attention for structural damage detections as post-processing algorithms. To accurately predict damages with ML, large amounts of structural response data are collected from a series dense sensor attached on the structure. Therefore, the damage diagnosis requires high computationally efforts. To address such issue, this paper presents a revolutionary approach utilizing image-based pre-trained convolutional neural network (CNN) to detect bridge damage locations and severities. In our research, scalograms from wavelet transform are adopted to convert structure acceleration data into image data. Compared with the traditional frequency analysis, which is derived from Fourier transform, the new method maintains both spatial and temporal information from the original structural behaviors. To generate CNN learning features, six channels of acceleration data are gathered from six strategically selected points of a finite element (FE) bridge model. Two pretrained CNN, AlexNet and Resnet, are selected to conduct transfer machine learning for higher training efficiency. The performances of the proposed method are assessed with various damage scenarios. The prediction accuracies of AlexNet and Resnet are 98% and 100%, respectively.

Presenting Author: Xi Song University of Hawaii at Manoa