Structural damage detection of simple beam subjected to a moving load using Deep Learning
Using Deep Learning of artificial intelligence(AI)for structural damage detection is a new direction in the field of structure health monitoring.The data generated by the structural health monitoring system is very complex.How to construct effective features to respond to the damage information contained in the huge data is a challenge.Deep Learning is very suitable for solving this problem because it can automatically learn the implicit feature information in the data in an unsupervised approach.In this paper,based on a high-level neural networks application programming interface(API)called Keras,the autoencoder is constructed to train the samples of acceleration responses,simulated using ANSYS software,of simple supported beam subjected to a moving load to extract the feature vectors.Finally,the damage index calculated by the feature vectors can be compared each other to judge the damage existence and location of the beam.The results show that the proposed method has the advantages of high accuracy for damage localization,feasibility and simplicity of the operation.
Damage detection Deep Learning Autoencoder Moving load Acceleration responses
Yuxin Wang Xirui Ma Zhenhua Nie Hongwei Ma
School of Mechanics and Construction Engineering,Jinan University,Guangzhou,China Dongguan University of Technology,Dongguan,China School of Mechanics and Construction Engineering,Jinan University,Guangzhou,China;Dongguan Universit
国际会议
The 7th World Conference on Structural Control and Monitoring(7WCSCM)(第七届结构控制与监测世界大会)
青岛
英文
2414-2420
2018-07-22（万方平台首次上网日期，不代表论文的发表时间）