Research On Bridge Damage Identification Based On CNN&LSTM

Traditional machine learning methods have rich research results in bridge damage recognition,and deep neural networks based on deep learning have great advantages over traditional machine learning methods.However,research on bridge damage identification based on deep neural networks is still relatively small,so deep neural networks such as convolutional neural networks and recurrent neural networks are selected for bridge damage identification research.The main research contents are as follows:1.Based on the finite element model of a cable-stayed bridge,the CEEMDAN method and the Hilbert transform method are used to extract the damage characteristics of the acceleration response of the finite element model.Finally,based on the extracted damage characteristics,a bridge damage sample library with different damage characteristics is constructed;2.Constructed a deep neural network model combined with CNN & LSTM,applied the built deep neural network model to the finite element damage sample library for training and testing,and at the same time,the combination feature was obtained as the optimal damage feature,and finally the optimal damage feature applied to three deep neural network models of CNN,LSTM,and MLP,the results show that the damage sample library constructed using the combined features can better reflect the damage status of the bridge,and the built deep neural network model can also identify the bridge damage more accurately Status,has great potential in practical applications;3.Based on the measured data of the shaking table test of a cable-stayed bridge,a bridge damage sample library with optimal damage characteristics was constructed,and then the deep neural network model combined with CNN & LSTM was applied to the measured damage sample library.Finally,the three comparisons of CNN,LSTM,and MLP were compared.The performance of a deep neural network model in the measured damage sample library.The results show that: the combined CNN & LSTM can more accurately identify the damage status of the bridge model of the cable-stayed bridge shaking table,which has great practical application value.