Cable Damage Identification of Tied-Arch Bridge Using Convolutional Neural Network

In the field of structural health monitoring, damage detection has been commonly carried out based on the modal properties and the engineering features related to the model. However, modal properties real-world structures are often affected by numerous uncertain factors, and the extracted features are also subjected to various errors, which makes the pattern recognition for damage detection still challenging. In this article, a damage detection method using the convolutional neural network (CNN) is presented for automated operation using raw measurement data without complex procedure for feature extraction. A CNN is a kind of deep neural network which typically consists of convolution, pooling, and fully-connected layers. A numerical simulation study was carried out for damage detection in cable hangers of a tied-arch bridge using ambient wind vibration data. Fourier amplitude spectra (FAS) of acceleration responses at cable anchorage points on the bridge deck are arranged as a matrix, which is used as the input to the CNN. Numerical results show that the current CNN using FAS data can detect both damage’ locations and extent accurately. Robustness of the present CNN has been proven under various observational noise levels and wind speeds.