基于深度学习的空间网架结构健康检测研究  

Research on health detection of space grid structure based on deep learning

作　　者：刘平[1] 邢广志 李雯芮 孙凯 LIU Ping;XIN Guangzhi;LI Wenrui;SUN Kai(School of Civil Engineering and Architecture,Jiangsu University of Science and Technology,Zhenjiang 212100,China)

机构地区：[1]江苏科技大学土木工程与建筑学院,镇江212100

出　　处：《江苏科技大学学报（自然科学版）》2023年第2期114-118,共5页Journal of Jiangsu University of Science and Technology:Natural Science Edition

基　　金：国家重点研发设计课题(2017YFC0821206);武警海警学院院级科研课题(YJKY06);国家自然科学基金资助项目(61871244)。

摘　　要：随着空间网架结构的普及,结构的健康检测是一个亟需解决的问题.根据网架结构中杆件分布规律,传力明确等特点,设计一个29个节点,76根杆件的网架模型.对模型分别在完好状态与部分杆件损伤时采集节点响应时程曲线,并将曲线时程信息作为输入向量,构造损伤信息矩阵,采用深度学习算法分步建立深度学习网络,并进行训练.通过建立精简化的网络,对单杆与多杆损伤工况进行损伤位置和损伤程度的识别.结果表明:建立的网络可以很好地识别结构的健康状态,该网络对实际应用具有参考价值.With the popularization of space grid structure,the health detection of structure is an urgent problem to be solved.According to the distribution law of members and the characteristics of clear force transmission,a grid model with 29 nodes and 76 members is designed.For the model,the response time-history curves of nodes are collected when the model is in good condition and some members are damaged,and the information of curve time-history is taken as the input vector to construct the damage information matrix.The deep learning network is established step by step by using the deep learning algorithm and trained.Through the establishment of a simplified network,the damage location and damage degree of single-and multi-bar are identified.The results show that the established network can identify the health state of the structure well,and it has reference value for application.

关 键 词：深度学习 空间网架结构 健康检测 损伤指标 简谐荷载 

分 类 号：TU391[建筑科学—结构工程]