Article(id=1228653711560470529, tenantId=1146029695717560320, journalId=1225147924628267009, issueId=1228653708687377017, articleNumber=null, orderNo=null, doi=10.16385/j.cnki.issn.1004-4523.2024.11.012, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1681401600000, receivedDateStr=2023-04-14, revisedDate=1686931200000, revisedDateStr=2023-06-17, acceptedDate=null, acceptedDateStr=null, onlineDate=1770863472117, onlineDateStr=2026-02-12, pubDate=1732723200000, pubDateStr=2024-11-28, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1770863472117, onlineIssueDateStr=2026-02-12, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1770863472117, creator=13701087609, updateTime=1770863472117, updator=13701087609, issue=Issue{id=1228653708687377017, tenantId=1146029695717560320, journalId=1225147924628267009, year='2024', volume='37', issue='11', pageStart='1803', pageEnd='1992', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1770863471433, creator=13701087609, updateTime=1770863902026, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1228655514792427773, tenantId=1146029695717560320, journalId=1225147924628267009, issueId=1228653708687377017, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1228655514792427774, tenantId=1146029695717560320, journalId=1225147924628267009, issueId=1228653708687377017, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=1917, endPage=1924, ext={EN=ArticleExt(id=1228653711778574340, articleId=1228653711560470529, tenantId=1146029695717560320, journalId=1225147924628267009, language=EN, title=Structural damage identification via a deep belief memory network, columnId=null, journalTitle=Journal of Vibration Engineering, columnName=null, runingTitle=null, highlight=null, articleAbstract=

Extracting sensitive damage features from structural response signals is crucial for damage identification methods based on pattern classification. To this end,a hybrid network that combines a deep belief networks (DBN) and a long-short term memory (LSTM) network is proposed through a hybrid learning mechanism to utilize the merits of both networks in the aspects of extracting high-order abstract features and considering data sequence correlations. First,transmissibility data from response signals are sequentially input into the DBN to achieve the initial data compression and feature extraction,reducing the redundant information in the responses. Then,the extracted feature sequences are input into the LSTM network to consider the correlation between the different responses for acquiring the relevant sensitive damage features. Finally,a classification layer with the Softmax function is used to classify the features output by the LSTM network. Thereby,different structural damage patterns can be identified. The damage identification results on a three-dimensional experimental steel frame demonstrate that the hybrid learning mechanism can better train the network parameters,and the fine-tuning on the whole hybrid network contributes to the subsequent damage feature classification. Under the pollution of numerical or measured noises,the hybrid network can still effectively perform the data compression,feature extraction and classification. The various damage scenarios of the experimental frame are well identified.

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=Sheng-En FANG, Yang LIU), CN=ArticleExt(id=1228653737388995313, articleId=1228653711560470529, tenantId=1146029695717560320, journalId=1225147924628267009, language=CN, title=结合深度信念记忆网络的结构损伤识别, columnId=0, journalTitle=振动工程学报, columnName=, runingTitle=null, highlight=null, articleAbstract=

从结构响应信号中挖掘敏感损伤特征是基于模式分类的损伤识别方法的关键。为此,将深度信念网络和长短期记忆网络进行混合组网,通过混合学习机制有机结合了两种网络在高阶抽象特征提取和考虑数据序列相关性上的优点。将响应信号传递比值输入深度信念网络,实现初步数据压缩和特征提取,以减少响应中的冗余信息;将特征序列依次输入长短期记忆网络,以考虑响应间的相关性并获取敏感损伤特征;利用Softmax分类层对长短期记忆网络输出的特征进行分类,实现对不同结构损伤模式的识别。三维试验钢框架的损伤识别结果表明:混合学习机制能更好地训练网络参数,整体微调后更有利于后续的损伤特征分类;混合组网方式在包含数值或实测噪声的情况下仍可以有效进行数据压缩、特征提取和分类,准确识别了试验框架的多种损伤工况。

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=HrT73ERa5wHhlOH/X1D16g==, magXml=3RbnIzlzdaetFFg+RZZiYQ==, pdfUrl=null, pdf=G/8TN9Izlw5TkT2Dij/tug==, pdfFileSize=2767570, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=khrqIRjDbszIxN23iiKDpg==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=P57VC44zcpM1UhKhBoLQkg==, mapNumber=null, authorCompany=null, fund=null, authors=

方圣恩(1980—),男,博士,教授,博士生导师。E-mail:

, authorsList=方圣恩, 刘洋)}, authors=[Author(id=1228653737883923210, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=shengen.fang@fzu.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1228653737980392206, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, authorId=1228653737883923210, language=EN, stringName=Sheng-En FANG, firstName=Sheng-En, middleName=null, lastName=FANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, address=1College of Civil Engineering,Fuzhou University,Fuzhou 350108,China
2National & Local Joint Engineering Research Center for Seismic and Disaster Informatization of Civil Engineering, Fuzhou University,Fuzhou 350108,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1228653738055889683, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, authorId=1228653737883923210, language=CN, stringName=方圣恩, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, address=1福州大学土木工程学院,福建 福州 350108
2福州大学土木工程防震减灾信息化国家地方联合工程研究中心,福建 福州 350108, bio={"content":"

方圣恩(1980—),男,博士,教授,博士生导师。E-mail:

"}, bioImg=null, bioContent=

方圣恩(1980—),男,博士,教授,博士生导师。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1228653737653236472, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, xref=1, ext=[AuthorCompanyExt(id=1228653737661625081, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, companyId=1228653737653236472, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Civil Engineering,Fuzhou University,Fuzhou 350108,China), AuthorCompanyExt(id=1228653737674207994, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, companyId=1228653737653236472, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学土木工程学院,福建 福州 350108)]), AuthorCompany(id=1228653737745511167, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, xref=2, ext=[AuthorCompanyExt(id=1228653737758094080, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, companyId=1228653737745511167, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2National & Local Joint Engineering Research Center for Seismic and Disaster Informatization of Civil Engineering, Fuzhou University,Fuzhou 350108,China), AuthorCompanyExt(id=1228653737766482689, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, companyId=1228653737745511167, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2福州大学土木工程防震减灾信息化国家地方联合工程研究中心,福建 福州 350108)])]), Author(id=1228653738169135894, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1228653738273993501, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, authorId=1228653738169135894, language=EN, stringName=Yang LIU, firstName=Yang, middleName=null, lastName=LIU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1College of Civil Engineering,Fuzhou University,Fuzhou 350108,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1228653738353685281, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, authorId=1228653738169135894, language=CN, stringName=刘洋, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1福州大学土木工程学院,福建 福州 350108, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1228653737653236472, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, xref=1, ext=[AuthorCompanyExt(id=1228653737661625081, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, companyId=1228653737653236472, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Civil Engineering,Fuzhou University,Fuzhou 350108,China), AuthorCompanyExt(id=1228653737674207994, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, companyId=1228653737653236472, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学土木工程学院,福建 福州 350108)])])], keywords=[Keyword(id=1228653738458542885, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, orderNo=1, keyword=damage identification), Keyword(id=1228653738508874536, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, orderNo=2, keyword=frame structure), Keyword(id=1228653738580177707, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, orderNo=3, keyword=deep belief network), Keyword(id=1228653738651480878, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, orderNo=4, keyword=long short-term memory network), Keyword(id=1228653740023018291, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, orderNo=5, keyword=hybrid learning mechanism), Keyword(id=1228653740123681589, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, orderNo=1, keyword=损伤识别), Keyword(id=1228653740232733499, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, orderNo=2, keyword=框架结构), Keyword(id=1228653740341785405, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, orderNo=3, keyword=深度信念网络), Keyword(id=1228653740413088575, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, orderNo=4, keyword=长短期记忆网络), Keyword(id=1228653740526334788, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, orderNo=5, keyword=混合学习机制)], refs=[Reference(id=1228653743135191961, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2020, volume=146, issue=5, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=SUN L M, SHANG Z Q, XIA Y, journalName=Journal of Structural Engineering, refType=null, unstructuredReference=SUN L MSHANG Z QXIA Y, et al. Review of bridge structural health monitoring aided by big data and artificial intelligence: from condition assessment to damage detection[J]. Journal of Structural Engineering2020146(5): 04020073., articleTitle=Review of bridge structural health monitoring aided by big data and artificial intelligence: from condition assessment to damage detection, refAbstract=null), Reference(id=1228653744523506592, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2008, volume=313, issue=3-5, pageStart=544, pageEnd=559, url=null, language=null, rfNumber=[2], rfOrder=1, authorNames=FANG S E, PERERA R, DE ROECK G, journalName=Journal of Sound and Vibration, refType=null, unstructuredReference=FANG S EPERERA RDE ROECK G. Damage identification of a reinforced concrete frame by finite element model updating using damage parameterization[J]. Journal of Sound and Vibration2008313(3-5):544-559., articleTitle=Damage identification of a reinforced concrete frame by finite element model updating using damage parameterization, refAbstract=null), Reference(id=1228653744628364198, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2011, volume=20, issue=11, pageStart=115009, pageEnd=null, url=null, language=null, rfNumber=[3], rfOrder=2, authorNames=REN W X, LIN Y Q, FANG S E, journalName=Smart Materials and Structures, refType=null, unstructuredReference=REN W XLIN Y QFANG S E. Structural damage detection based on stochastic subspace identification and statistical pattern recognition: Ⅰ. theory[J]. Smart Materials and Structures201120(11): 115009., articleTitle=Structural damage detection based on stochastic subspace identification and statistical pattern recognition: Ⅰ. theory, refAbstract=null), Reference(id=1228653744699667371, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2019, volume=32, issue=11, pageStart=1, pageEnd=20, url=null, language=null, rfNumber=[4], rfOrder=3, authorNames=孙利民, 尚志强, 夏烨, journalName=中国公路学报, refType=null, unstructuredReference=孙利民, 尚志强, 夏烨. 大数据背景下的桥梁结构健康监测研究现状与展望[J]. 中国公路学报201932(11): 1-20., articleTitle=大数据背景下的桥梁结构健康监测研究现状与展望, refAbstract=null), Reference(id=1228653744791942064, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2019, volume=32, issue=11, pageStart=1, pageEnd=20, url=null, language=null, rfNumber=[4], rfOrder=4, authorNames=SUN Limin, SHANG Zhiqiang, XIA Ye, journalName=China Journal of Highway and Transport, refType=null, unstructuredReference=SUN LiminSHANG ZhiqiangXIA Ye. Development and prospect of bridge structural health monitoring in the context of big data[J]. China Journal of Highway and Transport201932(11): 1-20., articleTitle=Development and prospect of bridge structural health monitoring in the context of big data, refAbstract=null), Reference(id=1228653744880022452, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2018, volume=196, issue=null, pageStart=44, pageEnd=54, url=null, language=null, rfNumber=[5], rfOrder=5, authorNames=GOMES G F, MENDÉZ Y A D, ALEXANDRINO P D S L, journalName=Composite Structures, refType=null, unstructuredReference=GOMES G FMENDÉZ Y A DALEXANDRINO P D S L, et al. The use of intelligent computational tools for damage detection and identification with an emphasis on composites-a review[J]. Composite Structures2018196: 44-54., articleTitle=The use of intelligent computational tools for damage detection and identification with an emphasis on composites-a review, refAbstract=null), Reference(id=1228653744976491448, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2016, volume=39, issue=8, pageStart=1697, pageEnd=1716, url=null, language=null, rfNumber=[6], rfOrder=6, authorNames=焦李成, 杨淑媛, 刘芳, journalName=计算机学报, refType=null, unstructuredReference=焦李成, 杨淑媛, 刘芳, 等. 神经网络七十年:回顾与展望[J]. 计算机学报201639(8): 1697-1716., articleTitle=神经网络七十年:回顾与展望, refAbstract=null), Reference(id=1228653745051988925, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2016, volume=39, issue=8, pageStart=1697, pageEnd=1716, url=null, language=null, rfNumber=[6], rfOrder=7, authorNames=JIAO Licheng, YANG Shuyuan, LIU Fang, journalName=Chinese Journal of Computers, refType=null, unstructuredReference=JIAO LichengYANG ShuyuanLIU Fang, et al. Seventy years beyond neural networks: retrospect and prospect[J]. Chinese Journal of Computers201639(8):1697-1716., articleTitle=Seventy years beyond neural networks: retrospect and prospect, refAbstract=null), Reference(id=1228653745144263616, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2013, volume=14, issue=4, pageStart=367, pageEnd=377, url=null, language=null, rfNumber=[7], rfOrder=8, authorNames=HAKIM S J S, RAZAK H A, journalName=Steel & Composite Structures, refType=null, unstructuredReference=HAKIM S J SRAZAK H A. Structural damage detection of steel bridge girder using artificial neural networks and finite element models[J]. Steel & Composite Structures201314(4): 367-377., articleTitle=Structural damage detection of steel bridge girder using artificial neural networks and finite element models, refAbstract=null), Reference(id=1228653745244926915, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2018, volume=30, issue=8, pageStart=2509, pageEnd=2518, url=null, language=null, rfNumber=[8], rfOrder=9, authorNames=VAFAEI M, ALIH S C, journalName=Neural Computing and Applications, refType=null, unstructuredReference=VAFAEI MALIH S C. Adequacy of first mode shape differences for damage identification of cantilever structures using neural networks[J]. Neural Computing and Applications201830(8): 2509-2518., articleTitle=Adequacy of first mode shape differences for damage identification of cantilever structures using neural networks, refAbstract=null), Reference(id=1228653745328812999, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2007, volume=null, issue=null, pageStart=1435, pageEnd=1442, url=null, language=null, rfNumber=[9], rfOrder=10, authorNames=ZHANG X D, ZHANG Z G, LI X F, journalName=null, refType=null, unstructuredReference=ZHANG X DZHANG Z GLI X F, et al. Damage identification in cable-stayed bridge based on modal analysis and neural networks[C]//AIP Conference Proceedings. Golden, 2007: 1435-1442., articleTitle=Damage identification in cable-stayed bridge based on modal analysis and neural networks, refAbstract=null), Reference(id=1228653745425281995, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2015, volume=521, issue=7553, pageStart=436, pageEnd=444, url=null, language=null, rfNumber=[10], rfOrder=11, authorNames=LECUN Y, BENGIO Y, HINTON G, journalName=Nature, refType=null, unstructuredReference=LECUN YBENGIO YHINTON G. Deep learning[J]. Nature2015521(7553): 436-444., articleTitle=Deep learning, refAbstract=null), Reference(id=1228653745496585170, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2019, volume=23, issue=5, pageStart=507, pageEnd=520, url=null, language=null, rfNumber=[11], rfOrder=12, authorNames=DUAN Y F, CHEN Q Y, ZHANG H M, journalName=Smart Structures and Systems, refType=null, unstructuredReference=DUAN Y FCHEN Q YZHANG H M, et al. CNN-based damage identification method of tied-arch bridge using spatial-spectral information[J]. Smart Structures and Systems201923(5): 507-520., articleTitle=CNN-based damage identification method of tied-arch bridge using spatial-spectral information, refAbstract=null), Reference(id=1228653745572082647, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2020, volume=20, issue=12, pageStart=3429, pageEnd=null, url=null, language=null, rfNumber=[12], rfOrder=13, authorNames=PURUNCAJAS B, VIDAL Y, TUTIVÉN C, journalName=Sensors, refType=null, unstructuredReference=PURUNCAJAS BVIDAL YTUTIVÉN C. Vibration-response-only structural health monitoring for offshore wind turbine jacket foundations via convolutional neural networks[J]. Sensors202020(12): 3429., articleTitle=Vibration-response-only structural health monitoring for offshore wind turbine jacket foundations via convolutional neural networks, refAbstract=null), Reference(id=1228653745647580122, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[13], rfOrder=14, authorNames=邱锡鹏, journalName=神经网络与深度学习, refType=null, unstructuredReference=邱锡鹏. 神经网络与深度学习[M]. 北京:机械工业出版社, 2020., articleTitle=null, refAbstract=null), Reference(id=1228653745731466205, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[13], rfOrder=15, authorNames=QIU Xipeng, journalName=Neural Networks and Deep Learning, refType=null, unstructuredReference=QIU Xipeng. Neural Networks and Deep Learning[M]. Beijing: China Machine Press, 2020., articleTitle=null, refAbstract=null), Reference(id=1228653745823740897, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2019, volume=31, issue=7, pageStart=1235, pageEnd=1270, url=null, language=null, rfNumber=[14], rfOrder=16, authorNames=YU Y, SI X S, HU C H, journalName=Neural Computation, refType=null, unstructuredReference=YU YSI X SHU C H, et al. A review of recurrent neural networks: LSTM cells and network architectures[J]. Neural Computation201931(7): 1235-1270., articleTitle=A review of recurrent neural networks: LSTM cells and network architectures, refAbstract=null), Reference(id=1228653745920209894, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2022, volume=35, issue=null, pageStart=436, pageEnd=451, url=null, language=null, rfNumber=[15], rfOrder=17, authorNames=SONY S, GAMAGE S, SADHU A, journalName=Structures, refType=null, unstructuredReference=SONY SGAMAGE SSADHU A, et al. Vibration-based multiclass damage detection and localization using long short-term memory networks[J]. Structures202235: 436-451., articleTitle=Vibration-based multiclass damage detection and localization using long short-term memory networks, refAbstract=null), Reference(id=1228653746008290281, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2013, volume=20, issue=1, pageStart=107, pageEnd=120, url=null, language=null, rfNumber=[16], rfOrder=18, authorNames=AN Y H, OU J P, journalName=Structural Control and Health Monitoring, refType=null, unstructuredReference=AN Y HOU J P. Experimental and numerical studies on model updating method of damage severity identification utilizing four cost functions[J]. Structural Control and Health Monitoring201320(1): 107-120., articleTitle=Experimental and numerical studies on model updating method of damage severity identification utilizing four cost functions, refAbstract=null), Reference(id=1228653746100564972, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2021, volume=33, issue=null, pageStart=68, pageEnd=76, url=null, language=null, rfNumber=[17], rfOrder=19, authorNames=ZHANG Y F, ZHU J S, journalName=Structures, refType=null, unstructuredReference=ZHANG Y FZHU J S. Damage identification for bridge structures based on correlation of the bridge dynamic responses under vehicle load[J]. Structures202133: 68-76., articleTitle=Damage identification for bridge structures based on correlation of the bridge dynamic responses under vehicle load, refAbstract=null), Reference(id=1228653746197033970, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2013, volume=32, issue=14, pageStart=82, pageEnd=86, url=null, language=null, rfNumber=[18], rfOrder=20, authorNames=闫维明, 顾大鹏, 陈彦江, journalName=振动与冲击, refType=null, unstructuredReference=闫维明, 顾大鹏, 陈彦江, 等. 基于加速度响应相关性的结构损伤识别方法[J]. 振动与冲击201332(14): 82-86., articleTitle=基于加速度响应相关性的结构损伤识别方法, refAbstract=null), Reference(id=1228653746297697269, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2013, volume=32, issue=14, pageStart=82, pageEnd=86, url=null, language=null, rfNumber=[18], rfOrder=21, authorNames=YAN Weiming, GU Dapeng, CHEN Yanjiang, journalName=Journal of Vibration and Shock, refType=null, unstructuredReference=YAN WeimingGU DapengCHEN Yanjiang, et al. A method for structural damage detection based on correlation characteristic of acceleration response[J]. Journal of Vibration and Shock201332(14): 82-86., articleTitle=A method for structural damage detection based on correlation characteristic of acceleration response, refAbstract=null), Reference(id=1228653746431915003, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2013, volume=30, issue=12, pageStart=99, pageEnd=106, url=null, language=null, rfNumber=[19], rfOrder=22, authorNames=杨小森, 闫维明, 陈彦江, journalName=公路交通科技, refType=null, unstructuredReference=杨小森, 闫维明, 陈彦江, 等. 基于振动信号统计特征的损伤识别方法[J]. 公路交通科技201330(12): 99-106., articleTitle=基于振动信号统计特征的损伤识别方法, refAbstract=null), Reference(id=1228653746515801086, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2013, volume=30, issue=12, pageStart=99, pageEnd=106, url=null, language=null, rfNumber=[19], rfOrder=23, authorNames=YANG Xiaosen, YAN Weiming, CHEN Yanjiang, journalName=Journal of Highway and Transportation Research and Development, refType=null, unstructuredReference=YANG XiaosenYAN WeimingCHEN Yanjiang, et al. Damage detection method based on statistics characteristics of vibration signal[J]. Journal of Highway and Transportation Research and Development201330(12): 99-106., articleTitle=Damage detection method based on statistics characteristics of vibration signal, refAbstract=null), Reference(id=1228653746578714624, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2015, volume=null, issue=null, pageStart=2625, pageEnd=2634, url=null, language=null, rfNumber=[20], rfOrder=24, authorNames=DONAHUE J, HENDRICKS L A, GUADARRAMA S, journalName=null, refType=null, unstructuredReference=DONAHUE JHENDRICKS L AGUADARRAMA S, et al. Long-term recurrent convolutional networks for visual recognition and description[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Washington, 2015: 2625-2634., articleTitle=Long-term recurrent convolutional networks for visual recognition and description, refAbstract=null), Reference(id=1228653746641629187, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2020, volume=138, issue=null, pageStart=106587, pageEnd=null, url=null, language=null, rfNumber=[21], rfOrder=25, authorNames=LEI Y G, YANG B, JIANG X W, journalName=Mechanical Systems and Signal Processing, refType=null, unstructuredReference=LEI Y GYANG BJIANG X W, et al. Applications of machine learning to machine fault diagnosis: a review and roadmap[J]. Mechanical Systems and Signal Processing2020138: 106587., articleTitle=Applications of machine learning to machine fault diagnosis: a review and roadmap, refAbstract=null), Reference(id=1228653746717126663, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2020, volume=53, issue=8, pageStart=5929, pageEnd=5955, url=null, language=null, rfNumber=[22], rfOrder=26, authorNames=HOUDT G V, MOSQUERA C, NÁPOLES G, journalName=Artificial Intelligence Review, refType=null, unstructuredReference=HOUDT G VMOSQUERA CNÁPOLES G. A review on the long short-term memory model[J]. Artificial Intelligence Review202053(8): 5929-5955., articleTitle=A review on the long short-term memory model, refAbstract=null), Reference(id=1228653746788429834, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2019, volume=117, issue=null, pageStart=453, pageEnd=482, url=null, language=null, rfNumber=[23], rfOrder=27, authorNames=YAN W J, ZHAO M Y, SUN Q, journalName=Mechanical Systems and Signal Processing, refType=null, unstructuredReference=YAN W JZHAO M YSUN Q, et al. Transmissibility-based system identification for structural health monitoring: fundamentals, approaches, and applications[J]. Mechanical Systems and Signal Processing2019117: 453-482., articleTitle=Transmissibility-based system identification for structural health monitoring: fundamentals, approaches, and applications, refAbstract=null), Reference(id=1228653746859733005, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2013, volume=332, issue=4, pageStart=807, pageEnd=820, url=null, language=null, rfNumber=[24], rfOrder=28, authorNames=MERUANE V, journalName=Journal of Sound and Vibration, refType=null, unstructuredReference=MERUANE V. Model updating using antiresonant frequencies identified from transmissibility functions[J]. Journal of Sound and Vibration2013332(4): 807-820., articleTitle=Model updating using antiresonant frequencies identified from transmissibility functions, refAbstract=null), Reference(id=1228653746960396303, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2012, volume=27, issue=3, pageStart=202, pageEnd=217, url=null, language=null, rfNumber=[25], rfOrder=29, authorNames=YAN W J, REN W X, journalName=Computer-Aided Civil and Infrastructure Engineering, refType=null, unstructuredReference=YAN W JREN W X. Operational modal parameter identification from power spectrum density transmissibility[J]. Computer-Aided Civil and Infrastructure Engineering201227(3):202-217., articleTitle=Operational modal parameter identification from power spectrum density transmissibility, refAbstract=null), Reference(id=1228653747031699474, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2022, volume=29, issue=8, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[26], rfOrder=30, authorNames=JU H Y, MA T, LI W, journalName=Structural Control and Health Monitoring, refType=null, unstructuredReference=JU H YMA TLI W, et al. Pixelwise asphalt concrete pavement crack detection via deep learning-based semantic segmentation method[J]. Structural Control and Health Monitoring202229(8): e2974., articleTitle=Pixelwise asphalt concrete pavement crack detection via deep learning-based semantic segmentation method, refAbstract=null), Reference(id=1228653747128168469, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[27], rfOrder=31, authorNames=伊恩·古德费洛, 约书亚·本吉奥, 亚伦·库维尔, 赵申剑, 黎或君, 符天凡, journalName=深度学习, refType=null, unstructuredReference=伊恩·古德费洛, 约书亚·本吉奥, 亚伦·库维尔. 深度学习[M]. 赵申剑, 黎或君, 符天凡, 等译. 北京:人民邮电出版社, 2021., articleTitle=null, refAbstract=null), Reference(id=1228653747287552024, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[28], rfOrder=32, authorNames=方开泰, 刘民千, 覃红, journalName=均匀试验设计的理论和应用, refType=null, unstructuredReference=方开泰, 刘民千, 覃红, 等. 均匀试验设计的理论和应用[M]. 北京:科学出版社, 2018., articleTitle=null, refAbstract=null)], funds=[Fund(id=1228653742837396368, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, awardId=52178276, language=CN, fundingSource=国家自然科学基金资助项目(52178276), fundOrder=null, country=null), Fund(id=1228653742942253972, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, awardId=2021J01601, language=CN, fundingSource=福建省自然科学基金资助项目(2021J01601), fundOrder=null, country=null), Fund(id=1228653743017751446, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, awardId=2021-Y-084, language=CN, fundingSource=福州市科技计划项目(2021-Y-084), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1228653737653236472, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, xref=1, ext=[AuthorCompanyExt(id=1228653737661625081, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, companyId=1228653737653236472, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Civil Engineering,Fuzhou University,Fuzhou 350108,China), AuthorCompanyExt(id=1228653737674207994, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, companyId=1228653737653236472, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学土木工程学院,福建 福州 350108)]), AuthorCompany(id=1228653737745511167, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, xref=2, ext=[AuthorCompanyExt(id=1228653737758094080, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, companyId=1228653737745511167, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2National & Local Joint Engineering Research Center for Seismic and Disaster Informatization of Civil Engineering, Fuzhou University,Fuzhou 350108,China), AuthorCompanyExt(id=1228653737766482689, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, companyId=1228653737745511167, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2福州大学土木工程防震减灾信息化国家地方联合工程研究中心,福建 福州 350108)])], figs=[ArticleFig(id=1228653740643775303, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, label=Fig.1, caption=Schematic diagram of a DBN topology, figureFileSmall=c9bXVacTdB5L1d070yIw3g==, figureFileBig=khrqIRjDbszIxN23iiKDpg==, tableContent=null), ArticleFig(id=1228653740731855690, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, label=图1, caption=DBN拓扑示意图, figureFileSmall=c9bXVacTdB5L1d070yIw3g==, figureFileBig=khrqIRjDbszIxN23iiKDpg==, tableContent=null), ArticleFig(id=1228653740970931025, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, label=Fig.2, caption=Schematic diagram of a LSTM basic unit, figureFileSmall=ogVnOqjAhrssVnzQOQoPWQ==, figureFileBig=l0Yw/z2mtGMLmi+bfHa21A==, tableContent=null), ArticleFig(id=1228653741038039891, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, label=图2, caption=LSTM基本单元示意, figureFileSmall=ogVnOqjAhrssVnzQOQoPWQ==, figureFileBig=l0Yw/z2mtGMLmi+bfHa21A==, tableContent=null), ArticleFig(id=1228653741109343061, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, label=Fig.3, caption=A DBN-LSTM network topology for damage identification, figureFileSmall=ESJJrxirV69eEgJ4VU2sWw==, figureFileBig=7ZZQG+4O1vVy4Mw+8aXbQg==, tableContent=null), ArticleFig(id=1228653741193229144, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, label=图3, caption=用于损伤识别的DBN-LSTM网络拓扑, figureFileSmall=ESJJrxirV69eEgJ4VU2sWw==, figureFileBig=7ZZQG+4O1vVy4Mw+8aXbQg==, tableContent=null), ArticleFig(id=1228653741285503834, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, label=Fig.4, caption=Damage identification procedure of a frame using the DBN-LSTM network, figureFileSmall=GNFXYCuP+lqQ3auafcgUTA==, figureFileBig=R16fslW3cGY0iGgRye56/Q==, tableContent=null), ArticleFig(id=1228653741386167132, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, label=图4, caption=基于DBN-LSTM网络的框架损伤识别流程, figureFileSmall=GNFXYCuP+lqQ3auafcgUTA==, figureFileBig=R16fslW3cGY0iGgRye56/Q==, tableContent=null), ArticleFig(id=1228653741465858912, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, label=Fig.5, caption=Test system of the experimental steel frame, figureFileSmall=hML99tlnNQusoKe/afcXLA==, figureFileBig=gpRTPnTOHjUnKIq6UTwHTg==, tableContent=null), ArticleFig(id=1228653741558133602, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, label=图5, caption=试验钢框架测试系统, figureFileSmall=hML99tlnNQusoKe/afcXLA==, figureFileBig=gpRTPnTOHjUnKIq6UTwHTg==, tableContent=null), ArticleFig(id=1228653741646213991, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, label=Fig.6, caption=Simulated and measured transmissibility functions of the steel frame, figureFileSmall=GevHpsj5vBMURr+hJo8jVQ==, figureFileBig=GvUzWN6E5eqqwbtGm3+Wsg==, tableContent=null), ArticleFig(id=1228653741734294378, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, label=图6, caption=钢框架数值和实测传递比函数, figureFileSmall=GevHpsj5vBMURr+hJo8jVQ==, figureFileBig=GvUzWN6E5eqqwbtGm3+Wsg==, tableContent=null), ArticleFig(id=1228653741851734895, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, label=Fig.7, caption=Training loss curves of the RBM, figureFileSmall=pAwrndIpz5AYdSbq0GUoOg==, figureFileBig=LITjS07h/RDbGKOu15S+KA==, tableContent=null), ArticleFig(id=1228653741964981107, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, label=图7, caption=RBM训练损失曲线, figureFileSmall=pAwrndIpz5AYdSbq0GUoOg==, figureFileBig=LITjS07h/RDbGKOu15S+KA==, tableContent=null), ArticleFig(id=1228653742040478581, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, label=Fig.8, caption=Fine-tuning results of DBN-LSTM network, figureFileSmall=3DocIb8GecF/WjXRUghx+w==, figureFileBig=x0/+vnWIAxV6q5HIB67A5g==, tableContent=null), ArticleFig(id=1228653742124364663, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, label=图8, caption=DBN-LSTM网络的微调结果, figureFileSmall=3DocIb8GecF/WjXRUghx+w==, figureFileBig=x0/+vnWIAxV6q5HIB67A5g==, tableContent=null), ArticleFig(id=1228653742204056441, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, label=Fig.9, caption=Numerical verification of DBN-LSTM network, figureFileSmall=gUM0U6kmYoowgpZ4vNdKiA==, figureFileBig=YrAOesCyRz/5Q9izK6ofKQ==, tableContent=null), ArticleFig(id=1228653742300525437, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, label=图9, caption=DBN-LSTM网络数值验证结果, figureFileSmall=gUM0U6kmYoowgpZ4vNdKiA==, figureFileBig=YrAOesCyRz/5Q9izK6ofKQ==, tableContent=null), ArticleFig(id=1228653742376022912, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, label=Fig.10, caption=Text verification of DBN-LSTM network, figureFileSmall=Fc1+clBBV4oimgMWUgWb+A==, figureFileBig=GMm00YdVWc7sN0FvBrEGuA==, tableContent=null), ArticleFig(id=1228653742464103301, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, label=图10, caption=DBN-LSTM网络试验验证, figureFileSmall=Fc1+clBBV4oimgMWUgWb+A==, figureFileBig=GMm00YdVWc7sN0FvBrEGuA==, tableContent=null), ArticleFig(id=1228653742631875465, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=EN, label=Tab.1, caption=

Sample divisions of the training set,validation set and testing set

, figureFileSmall=null, figureFileBig=null, tableContent=
损伤工况训练集验证集测试集合计
未损伤3000100010005000
单层损伤37500125001250062500
双层损伤225007500750037500
), ArticleFig(id=1228653742711567244, tenantId=1146029695717560320, journalId=1225147924628267009, articleId=1228653711560470529, language=CN, label=表1, caption=

训练集、验证集和测试集样本配比

, figureFileSmall=null, figureFileBig=null, tableContent=
损伤工况训练集验证集测试集合计
未损伤3000100010005000
单层损伤37500125001250062500
双层损伤225007500750037500
)], attaches=null, journal=Journal(id=1225147830491308032, delFlag=0, nameCn=振动工程学报, nameEn=Journal of Vibration Engineering, nameHistory1=null, nameHistory2=null, issn=1004-4523, eissn=null, cn=32-1349/TB, coden=null, periodic=0, language=CN, oaType=null, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=null, journalPrice=null, startedYear=null, abbrevIsoEn=Journal of Vibration Engineering, journalRemark=null, publicationField=null, createdTime=1770027604939, updatedTime=1770169610881, createdBy=18614031015, updatedBy=18614031015, firstLetterCn=J, firstLetterEn=J, subjectCode=Engineering, subjectName=null, subjectCodeEn=Engineering, subjectNameEn=null, picCn=null, picEn=null, jcr=null, cjcr=null, exts=[JournalExt(id=1225743346702925905, language=CN, name=振动工程学报, nameHistory1=null, nameHistory2=null, managedBy=中国科学技术协会, sponsoredBy=中国振动工程学会, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1770169587064, updatedTime=1770169587064, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://www.manuscripts.com.cn/zdgcxb, submissionEditorUrl=https://www.manuscripts.com.cn/zdgcxb, submissionReviewUrl=https://www.manuscripts.com.cn/zdgcxb, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1225743346765840466, language=EN, name=Journal of Vibration Engineering, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1770169587079, updatedTime=1770169587079, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://www.manuscripts.com.cn/zdgcxb, submissionEditorUrl=https://www.manuscripts.com.cn/zdgcxb, submissionReviewUrl=https://www.manuscripts.com.cn/zdgcxb, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1225147924628267009, websiteList=[Website(id=1225150618881404985, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1225147924628267009, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/zdgcxb/CN, language=CN, createTime=1770028269739, createBy=18614031015, updateTime=1770028293069, updateBy=18614031015, name=振动工程学报-中文, tplId=1146099689490845704, title=振动工程学报, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1225151164178673750, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150618881404985, code=articleTextType, value=kx, createTime=1770028399748, updateTime=1770028399748, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151164157702227, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150618881404985, code=banner, value=null, createTime=1770028399743, updateTime=1770028399743, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151164203839577, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150618881404985, code=grayFlag, value=0, createTime=1770028399754, updateTime=1770028399754, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151164145119314, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150618881404985, code=logo, value=https://castjournals.cast.org.cn/joweb/zdgcxb/EN/file/pic?fileId=L7mSU8YPwm66NWFMoTG4aQ==, createTime=1770028399740, updateTime=1770028399740, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151164212228187, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150618881404985, code=minRunFlag, value=0, createTime=1770028399756, updateTime=1770028399756, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151164170285141, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150618881404985, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/zdgcxb/CN/file/pic, createTime=1770028399746, updateTime=1770028399746, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151164208033882, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150618881404985, code=silenceFlag, value=0, createTime=1770028399755, updateTime=1770028399755, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151164166090836, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150618881404985, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1770028399745, updateTime=1770028399745, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151164187062359, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150618881404985, code=themeColor, value=null, createTime=1770028399750, updateTime=1770028399750, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151164195450968, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150618881404985, code=themeStyle, value=null, createTime=1770028399752, updateTime=1770028399752, creator=18614031015, updator=18614031015)]), Website(id=1225150619003039804, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1225147924628267009, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/zdgcxb/EN, language=EN, createTime=1770028269768, createBy=18614031015, updateTime=1770028309190, updateBy=18614031015, name=振动工程学报-英文, tplId=1146101810881728533, title=Journal of Vibration Engineering, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1225151193366835296, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150619003039804, code=articleTextType, value=kx, createTime=1770028406707, updateTime=1770028406707, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151193350058077, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150619003039804, code=banner, value=null, createTime=1770028406703, updateTime=1770028406703, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151193387806819, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150619003039804, code=grayFlag, value=0, createTime=1770028406712, updateTime=1770028406712, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151193341669468, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150619003039804, code=logo, value=https://castjournals.cast.org.cn/joweb/zdgcxb/EN/file/pic?fileId=L7mSU8YPwm66NWFMoTG4aQ==, createTime=1770028406701, updateTime=1770028406701, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151193400389733, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150619003039804, code=minRunFlag, value=0, createTime=1770028406715, updateTime=1770028406715, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151193362640991, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150619003039804, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/zdgcxb/EN/file/pic, createTime=1770028406706, updateTime=1770028406706, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151193392001124, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150619003039804, code=silenceFlag, value=0, createTime=1770028406713, updateTime=1770028406713, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151193354252382, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150619003039804, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1770028406704, updateTime=1770028406704, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151193371029601, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150619003039804, code=themeColor, value=null, createTime=1770028406708, updateTime=1770028406708, creator=18614031015, updator=18614031015), WebsiteProps(id=1225151193379418210, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1225150619003039804, code=themeStyle, value=null, createTime=1770028406710, updateTime=1770028406710, creator=18614031015, updator=18614031015)])], journalTitle=振动工程学报, weixinUrl=null, journalUrl=http://zdgcxb.csve.org.cn/, iacademicId=null, status=1, seqNo=null, journalTitleEn=Journal of Vibration Engineering, journalPhotoCn=null, journalPhotoEn=null, journalFirstLetter=J, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/zdgcxb/CN/10.16385/j.cnki.issn.1004-4523.2024.11.012, detailUrlEn=https://castjournals.cast.org.cn/joweb/zdgcxb/EN/10.16385/j.cnki.issn.1004-4523.2024.11.012, pdfUrlCn=https://castjournals.cast.org.cn/joweb/zdgcxb/CN/PDF/10.16385/j.cnki.issn.1004-4523.2024.11.012, pdfUrlEn=https://castjournals.cast.org.cn/joweb/zdgcxb/EN/PDF/10.16385/j.cnki.issn.1004-4523.2024.11.012, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
结合深度信念记忆网络的结构损伤识别
收藏切换
PDF下载
方圣恩 1, 2 , 刘洋 1
振动工程学报 | 2024,37(11): 1917-1924
收起
收藏切换
振动工程学报 | 2024, 37(11): 1917-1924
结合深度信念记忆网络的结构损伤识别
全屏
方圣恩1, 2 , 刘洋1
作者信息
  • 1福州大学土木工程学院,福建 福州 350108
  • 2福州大学土木工程防震减灾信息化国家地方联合工程研究中心,福建 福州 350108
  • 方圣恩(1980—),男,博士,教授,博士生导师。E-mail:

Structural damage identification via a deep belief memory network
Sheng-En FANG1, 2 , Yang LIU1
Affiliations
  • 1College of Civil Engineering,Fuzhou University,Fuzhou 350108,China
  • 2National & Local Joint Engineering Research Center for Seismic and Disaster Informatization of Civil Engineering, Fuzhou University,Fuzhou 350108,China
出版时间: 2024-11-28 doi: 10.16385/j.cnki.issn.1004-4523.2024.11.012
文章导航
收藏切换

从结构响应信号中挖掘敏感损伤特征是基于模式分类的损伤识别方法的关键。为此,将深度信念网络和长短期记忆网络进行混合组网,通过混合学习机制有机结合了两种网络在高阶抽象特征提取和考虑数据序列相关性上的优点。将响应信号传递比值输入深度信念网络,实现初步数据压缩和特征提取,以减少响应中的冗余信息;将特征序列依次输入长短期记忆网络,以考虑响应间的相关性并获取敏感损伤特征;利用Softmax分类层对长短期记忆网络输出的特征进行分类,实现对不同结构损伤模式的识别。三维试验钢框架的损伤识别结果表明:混合学习机制能更好地训练网络参数,整体微调后更有利于后续的损伤特征分类;混合组网方式在包含数值或实测噪声的情况下仍可以有效进行数据压缩、特征提取和分类,准确识别了试验框架的多种损伤工况。

损伤识别  /  框架结构  /  深度信念网络  /  长短期记忆网络  /  混合学习机制

Extracting sensitive damage features from structural response signals is crucial for damage identification methods based on pattern classification. To this end,a hybrid network that combines a deep belief networks (DBN) and a long-short term memory (LSTM) network is proposed through a hybrid learning mechanism to utilize the merits of both networks in the aspects of extracting high-order abstract features and considering data sequence correlations. First,transmissibility data from response signals are sequentially input into the DBN to achieve the initial data compression and feature extraction,reducing the redundant information in the responses. Then,the extracted feature sequences are input into the LSTM network to consider the correlation between the different responses for acquiring the relevant sensitive damage features. Finally,a classification layer with the Softmax function is used to classify the features output by the LSTM network. Thereby,different structural damage patterns can be identified. The damage identification results on a three-dimensional experimental steel frame demonstrate that the hybrid learning mechanism can better train the network parameters,and the fine-tuning on the whole hybrid network contributes to the subsequent damage feature classification. Under the pollution of numerical or measured noises,the hybrid network can still effectively perform the data compression,feature extraction and classification. The various damage scenarios of the experimental frame are well identified.

damage identification  /  frame structure  /  deep belief network  /  long short-term memory network  /  hybrid learning mechanism
方圣恩, 刘洋. 结合深度信念记忆网络的结构损伤识别. 振动工程学报, 2024 , 37 (11) : 1917 -1924 . DOI: 10.16385/j.cnki.issn.1004-4523.2024.11.012
Sheng-En FANG, Yang LIU. Structural damage identification via a deep belief memory network[J]. Journal of Vibration Engineering, 2024 , 37 (11) : 1917 -1924 . DOI: 10.16385/j.cnki.issn.1004-4523.2024.11.012
土木结构长期处于复杂服役环境,不可避免会产生和累积不同程度的损伤,导致结构性能不断退化,若不能及时发现损伤并采取有效措施,就可能引起灾难性的事故1。利用频率、振型等模态参数构建目标函数,通过反问题求解进行损伤识别2,或利用模式归类来判断结构是否发生损伤3,是研究中常用的两类方法。反问题求解过程如模型修正容易出现病态灵敏度矩阵或陷入局部最优,此时可以通过结构损伤特征的归类来实现有效的损伤判断。
通过损伤模式归类能判断土木结构是否发生损伤,属于模式识别问题的求解过程3,可以采用人工神经网络(Artificial Neural Networks,ANNs)、支持向量机和决策树等方法实现4。其中,ANNs能够通过参数学习来逼近损伤特征与标签(结构状态)之间复杂的非线性映射关系,特别适合用于模式识别问题,过去20年来已取得了较多的研究成果5
ANNs可分为浅层神经网络和深度神经网络(Deep Neural Networks,DNNs)两大类6。传统浅层网络的拓扑相对简单,通常只包含一二个隐藏层,无法挖掘高维响应信号中的敏感特征,损伤识别应用时常采用模态参数及其衍生参数作为损伤特征,如模态频率7、振型8和曲率9等,容易受环境、测试噪声和测点数量等因素的影响,在搜索空间中的离散性也会限制模式分类的处理能力。DNNs通过隐藏层堆叠的方式逐层提取原始输入信息的高阶抽象特征,可将其应用在语音识别、图像识别等领域10。用于结构损伤识别时,DNNs可以输入包含更多结构信息的振动响应信号,如频响函数、响应谱11和加速度响应12等,再利用多个隐藏层自动挖掘输入响应中包含的结构信息,减少冗余和不相关特征,得到对结构损伤敏感的特征,有利于分类层更好地进行特征分类,进而实现高效的损伤识别。
相较于卷积神经网络和栈式自编码器等层间全连接、层内无连接的DNNs,循环神经网络能够较好地考虑输入数据间的序列相关性,以获取更敏感的数据特征13。但实际运用时,若输入序列比较长,网络会存在梯度爆炸或消失问题,即长程依赖问题。而长短期记忆(Long Short⁃Term Memory,LSTM)网络通过门控机制对信息的累积速度进行控制,包括选择性地添加新的信息,以及选择性地忽略以前积累的信息,以此有效地解决长程依赖问题,因此可以更好地考虑输入数据序列间的相关性,获取数据的敏感特征14。SONY等15将加速度信号按时间序列输入LSTM网络,实现了一座看台模型的损伤定位,但仅考虑了加速度信号的时间相关性,未包含不同响应间的空间相关性。而实践中发现不同位置的响应存在一定关联1619。闫维明等18证明了加速度响应的相关特性包含了结构的模态信息,与损伤程度和位置相关,可作为结构损伤因子。杨小森等19发现无论是正弦激励还是白噪声激励,结构位移响应信号的统计特征(如相关系数、回归系数和协方差之比)都包含了结构的模态信息,可由此构建损伤指标。
全连接神经网络和循环神经网络的混合组网能有机结合不同类型网络的优点,可以利用全连接神经网络对原始输入信息进行初步数据压缩和特征提取,减少输入中的冗余信息,再通过循环神经网络融合前后输入的序列相关性,实现对输入数据更充分的特征挖掘。比如在图像分析领域,利用卷积神经网络提取视频每帧图像的特征并减少输入表示中的冗余,再按序列依次输入LSTM网络,以融合每帧图像特征的序列相关性,实现视频活动识别、图像标题生成和视频描述20
深度信念网络(Deep Belief Network,DBN)是一种全连接的深度神经网络,其隐藏层能够逐层提取原始输入信息中的高阶抽象特征,以减少输入中的冗余信息21。为此,本文在损伤识别问题上尝试将DBN与LSTM网络进行混合组网,构建深度信念记忆(DBN⁃LSTM)网络,采用无监督和有监督学习相结合的混合学习机制进行混合网的训练,快速获取合适的训练参数。损伤识别时,首先将每个响应信号按序列依次输入,利用DBN降低输入响应的维度和表示冗余,进行初步数据压缩和损伤特征提取;然后将特征按照响应信号的序列依次输入LSTM网络,充分考虑响应间存在的时空相关性(如不同响应间的统计特征),得到包含不同响应相关性的敏感损伤特征;最后利用Softmax分类层对LSTM网络输出的特征进行分类,准确识别结构的不同损伤工况。文末通过一榀三维框架模型验证了所提方法的可行性。
DBN是由多个受限波尔兹曼机(Restricted Boltzmann Machine,RBM)通过逐层贪婪学习策略得到的DNN21,每个隐藏层都能够基于前一个隐藏层提取的特征信息,进一步提取对于原始输入信息的高阶抽象特征,从而更好地完成对原始输入信息的压缩和特征提取。图1展示了由3个RBM堆栈而成的DBN拓扑,每个 RBM都包含可视层和隐藏层组成RBM1,再将作为,与组成RBM2。同样地,将作为,与组成RBM3。网络训练时,输入的数据依次逐层传递至,实现对原始输入信息的压缩和特征提取。
LSTM网络是一种循环神经网络,基本单元包含遗忘门、输入门和输出门,通过门控机制来控制信息的累积速度,包括有选择地加入新的信息,并有选择地遗忘之前累积的信息,可以很好地考虑输入数据序列的相关性,如图2所示22
遗忘门决定上一个记忆要保留的部分,即向量内元素为1说明全部保留,为0则全部遗忘;输入门决定要更新的部分,即内元素为1表示全部保留至本单元的记忆中,为0则不保留;输出门决定本单元的输出内元素为1,则说明经过tanh函数转换的全部作为输出,为0时则输出0。上述参数的表达式如下所示22
式中  分别为从输入向量到输入门、输出门、遗忘门和记忆的权重矩阵;分别为从中间输出到输入门、输出门、遗忘门和记忆的权重矩阵;分别为输入门、输出门、遗忘门和记忆的偏置项;表示向量中元素按位相乘;表示Sigmoid函数;表示tanh函数。
针对结构损伤模式识别问题,为实现更有效的数据压缩和特征提取,同时充分考虑不同测点响应间的相关性,进而实现更有效的损伤模式分类,本文采用DBN与LSTM混合组网的方式,构建了DBN⁃LSTM网络,其拓扑由输入层、特征提取层、分类层和输出层组成,如图3所示。输入层是整个网络的入口,为更好地考虑不同响应间的相关性,参考循环神经网络处理自然语言中“词向量”的表示方法,将响应信号(图3S1~Sn)按序列依次输入;特征提取层由DBN隐藏层和LSTM网络组合而成,是整个DBN⁃LSTM网络的核心,其中DBN对响应信号进行逐层提取,得到与输入相关的高阶抽象特征(图3F1~Fn),减少输入中表示的冗余信息,再将特征依次输入LSTM网络,进行充分融合,形成新的包含响应间相关特性的敏感特征,并输出到后续分类层;分类层以具有多类别分类功能的Softmax函数作为激活函数,能够映射损伤特征与结构状态之间的非线性关系;分类结果由输出层输出。
结构损伤识别往往依赖于激励⁃响应信号分析所获取的动力特征,但对处于环境激励下的实际土木工程结构而言,由于激励力未知,因此无法获取频响函数。而传递比函数反映的是系统输出⁃输出(响应⁃响应)间的关系,更适合于仅能测量响应时的土木结构动力分析以及损伤识别问题的求解23,其不涉及激励的测量,可采用响应谱计算,且对加速度、速度和位移响应都适用,因此应用上更加方便2425
式中  表示在点激励系统时响应点ij之间的传递比函数;表示i点的响应谱,表示j点的响应谱;的共轭复数。
激励力未知时,可采用进行计算,即以响应测点为参考点,计算响应点ij之间的传递比函数。此时表示i点的响应谱,表示j点的响应谱;表示p点的响应谱,共轭复数。
采用DNNs进行分析时往往需要大量的样本,对实际土木结构来说通常难以实现。此时可通过有限元模拟获取,转换为结构的频响函数进行计算23
式中  表示在k点激励系统时响应点i的频响函数;表示在k点激励系统时响应点j的频响函数。
DBN⁃LSTM网络训练样本中的值也通过有限元分析获取。样本集包含训练集、验证集和测试集三种,常用占比为3∶1∶126,其中训练集用于训练网络中的权重和偏置参数,验证集用于训练学习率、训练次数、训练批次、隐藏层数及相应的神经元数等超参数和检验模型性能,测试集用于检验网络的泛化能力27
为了尽可能涵盖结构可能的损伤工况,神经网络样本集需要综合考虑各种损伤模式和程度,使得样本数目变得庞大,增加了网络的训练时间和计算成本。为此,本文利用均匀设计28来减少样本量,将损伤位置作为试验因素,损伤程度作为试验水平。具体到框架结构的损伤识别上,单层损伤工况可采用单因素、多水平设计,多损伤工况则采用多因素、多水平的均匀设计。另外,实测中通常包含环境及测试噪声,为了增强DBN⁃LSTM网络对噪声的鲁棒性,在数值样本中加入了高斯白噪声,使得训练与测试用的值均含有噪声。
要说明的是,在频响数据样本处理上,网络训练时将DBN提取的传递比函数的高阶抽象特征(图3F1~Fn)依次输入LSTM网络,利用其可以很好地考虑序列相关性的优势得到损伤敏感特征,响应间相关性的信息都包含在相应LSTM单元的记忆和输出中。
以框架结构为例,基于DBN⁃LSTM网络的损伤识别流程如图4所示。首先,建立试验框架的有限元模型,通过有限元分析获取包含不同工况传递比数值的训练集、验证集和测试集样本;其次,搭建DBN⁃LSTM网络拓扑,设置输入层和输出层神经元的数目、LSTM单元数,并初设DBN隐藏层数及相应神经元个数、分类层神经元个数,同时设置学习率和训练批次数目等参数;接着,通过训练和验证过程优化调整DBN⁃LSTM网络的拓扑和权重等参数,其中训练集用于确定网络权重、偏置等参数,验证集用于确定网络的学习率、训练次数、训练批次、隐藏层数及相应的神经元数等超参数;随后,利用测试集来评价DBN⁃LSTM网络对未知样本的预测准确度;最后,将试验框架的实测传递比值代入DBN⁃LSTM网络,实现对框架的损伤识别。
具体地,网络训练过程包括预训练和网络微调两个步骤,其中预训练是对所有RBM进行逐个训练,属于无监督学习过程,能够得到与训练集相关的隐藏层参数(权重和偏置)初始值,加速网络的收敛;微调则是将DBN隐藏层、LSTM网络和分类层进行同时训练,进一步调整预训练后的网络参数初始值,使整个DBN⁃LSTM网络对训练样本有更好的拟合,对未知样本也有更好的预测效果。网络验证过程则是将验证集代入前述训练后的DBN⁃LSTM网络,通过计算损失和准确率来判断网络是否符合要求。此处准确率指所有预测结果中正确预测数量的样本占比,微调训练损失则采用交叉熵损失函数计算13
式中  表示真实值;表示预测值;n为类别数。
若损失和准确率满足要求,则保存当前网络的各参数(权重、偏置和超参数);反之,则对网络超参数(隐藏层个数、神经元个数、训练批次数目和学习率等)进行调整,然后重新预训练和微调,直到代入验证集后,DBN⁃LSTM网络的输出损失和预测准确率符合要求,此时认为网络训练完成。
采用一榀三维5层试验钢框架模型验证所提出方法的可行性,如图5所示。框架全高1250 mm,每层高250 mm。梁、板、柱钢材的实测弹性模量均值为184.7 GPa,密度为。采用锤击法激励框架,激励点位于第五层梁的中心,加速度传感器布设于每层楼板的中心,通过动态测试系统采集各测点的加速度时程数据,采样频率为100 Hz,谱线数为1600,每个传递比函数有1600个数值。
试验中通过减少整层4根柱的截面宽度(由初始30 mm减至25 mm,换算为刚度降低16.7%)来模拟损伤,共测试了单层损伤(L2,L4)和双层损伤(L2+L4)3种工况。为生成DBN⁃LSTM网络所需的训练样本,建立了框架的有限元模型(见图4),几何尺寸与试验模型相同,材料特性采用实测数据。梁、板、柱均采用壳单元模拟,网格划分后共包含1684个壳单元。
如前所述,框架损伤识别所用的DBN⁃LSTM网络包含输入层、特征提取层、分类层和输出层。对框架前2阶频率的传递比数据(0~11.1 Hz频段,包含453个传递比值)进行压缩和特征提取,故网络输入层有453个神经元;LSTM网络部分有4个LSTM单元,与输入的响应数目相同;分类层和输出层神经元个数均为9,与预设的损伤模拟工况数目相同;初设训练批次数目、学习率、每个RBM的预训练次数、微调训练次数、DBN隐藏层个数及相应的神经元个数等超参数。
如前所述,DBN⁃LSTM网络的训练集、验证集和测试集由框架有限元模型计算得到,包含各种不同损伤位置和程度的工况。损伤通过减小柱的弹性模量来模拟,样本集包含单层和双层损伤工况:单层损伤有L1~L5层分别损伤,共5个模拟工况,损伤程度在2%~50%范围内均匀分布,步长2%;双层损伤以间隔为主,有L1+L3,L2+L4,L3+L5损伤3种模拟工况,损伤程度根据均匀设计表进行取值,损伤程度也为2%~50%。未损伤框架作为独立工况加入样本集,因此一共9个模拟工况,对应输出层9个神经元。此外,为模拟实际情况,增加DBN⁃LSTM网络的鲁棒性,本文利用信号与标准正态分布的随机矩阵叠加的方式,给各工况的传递比函数值添加1%~10%的高斯白噪声,如图6所示。
如2.2节所述,训练集、验证集、测试集的样本数比例为3∶1∶1。对于未损伤工况,均匀抽取 40%样本,验证集和测试集各占一半,剩余 60%的样本作为训练集;对于单层损伤工况,均匀选取损伤程度为4%,14%,24%,34%,44%的样本作为测试集,损伤程度为8%,18%,28%,38%,48%的样本作为验证集,剩余的样本作为训练集;对于双层损伤工况,取均匀设计表中水平为2,7,12,17,22的样本作为测试集,水平为4,9,14,19,24的样本作为验证集,其他的样本作为训练集。样本集的总数为105000个,各样本集的分配如表1所示,1个样本包含由5个响应测试点计算的4个传递比函数。
本文基于TensorFlow平台搭建DBN⁃LSTM网络,预训练采用对比散度学习算法,微调训练采用Adam优化算法13。经过预训练、微调和验证,确定网络的训练批次数目为100、学习率为0.0001,每个RBM隐藏层的预训练次数为500,微调训练次数为200,DBN隐藏层个数为3,各层神经元个数分别依次为300,150,50;最后一个LSTM单元的输出为3个神经元。
3个RBM训练的损失曲线如图7所示。每个RBM的训练损失在开始阶段即大幅下降,第100轮后基本趋于稳定,说明RBM训练的参数(权重和偏置)学习主要在前期完成。
DBN⁃LSTM网络的整体微调结果如图8所示。训练集的损失值随着训练次数的增加逐渐趋近于0,准确率则逐渐稳定至99.98%,说明网络很好地学习了如何对训练集数据进行初步数据压缩并融合响应间相关性,以得到敏感损伤特征和分类。此外,训练集、验证集的损失曲线和准确率曲线在整个微调过程中基本保持一致,说明网络能对训练集以外的样本进行准确分类,不存在过拟合问题,可以保存此时的隐藏层个数、神经元个数、训练批次数目和学习率等超参数。图8(b),(c)分别为训练集和验证集的损伤特征在三维空间中的分布,特征由LSTM的最后1个单元输出,3个神经元数值依次对应XYZ 的坐标值,可见9种工况的特征可以被很好地区分,有利于Softmax分类层进行更有效的分类。
最后,将测试集代入经验证的DBN⁃LSTM网络中,得到的损伤识别准确率为99.97%,如图9所示。图9(a)为分类结果的混淆矩阵,可见9种工况中只有未损伤、单层损伤L3和L5出现了极个别的误判(比如未损伤1000个样本中,2个样本误判),其他工况都达到了100%的识别准确率。图9(b)为未损伤、单层损伤、双层损伤在不同噪声程度下的识别准确率,可见噪声程度无论是0%~3%,3%~7%还是7%~10%,DBN⁃LSTM网络都能够进行准确的识别,表现出良好的抗噪性。
针对单层损伤L2,L4和双层损伤L2+L4三种工况,进行了实测效果验证。将实测传递比数据输入DBN⁃LSTM网络中,预测结果如图10所示。图中横坐标表示工况,纵坐标表示识别结果属于某工况的概率。由图可见,所提方法准确识别了单层损伤L2、单层损伤L4及双层损伤L2+L4工况。
为提高复杂结构损伤模式分类的准确率,本文提出一种基于深度信念记忆(DBN⁃LSTM)网络的损伤识别方法,有机结合了DBN在减少输入表示中的冗余和LSTM在考虑输入表示的序列相关性上的优点,实现对传递比函数数据的充分挖掘,得到敏感特征。三维试验框架研究结果表明:
(1)DBN和LSTM混合网络良好融合了全连接神经网络提取输入高阶抽象特征和循环神经网络考虑数据序列相关性的优点,有效实现了对框架结构敏感损伤特征的准确分类。
(2)DBN⁃LSTM网络采用的无监督和有监督的混合学习机制能更好地获取合适的权重和偏置等参数,同时经过整体微调的网络更有利于后续的损伤特征分类。
(3)DBN⁃LSTM网络在传递比函数值包含噪声的情况下仍可以有效进行数据压缩、特征提取和分类,准确识别了框架的多种损伤工况,体现出良好的抗噪性。实测的试验传递比数据作为输入时,网络仍保持了优异的损伤分类和识别性能。
  • 国家自然科学基金资助项目(52178276)
  • 福建省自然科学基金资助项目(2021J01601)
  • 福州市科技计划项目(2021-Y-084)
参考文献 引证文献
排序方式:
[1]
SUN L MSHANG Z QXIA Y, et al. Review of bridge structural health monitoring aided by big data and artificial intelligence: from condition assessment to damage detection[J]. Journal of Structural Engineering2020146(5): 04020073.
[2]
FANG S EPERERA RDE ROECK G. Damage identification of a reinforced concrete frame by finite element model updating using damage parameterization[J]. Journal of Sound and Vibration2008313(3-5):544-559.
[3]
REN W XLIN Y QFANG S E. Structural damage detection based on stochastic subspace identification and statistical pattern recognition: Ⅰ. theory[J]. Smart Materials and Structures201120(11): 115009.
[4]
孙利民, 尚志强, 夏烨. 大数据背景下的桥梁结构健康监测研究现状与展望[J]. 中国公路学报201932(11): 1-20.
SUN LiminSHANG ZhiqiangXIA Ye. Development and prospect of bridge structural health monitoring in the context of big data[J]. China Journal of Highway and Transport201932(11): 1-20.
[5]
GOMES G FMENDÉZ Y A DALEXANDRINO P D S L, et al. The use of intelligent computational tools for damage detection and identification with an emphasis on composites-a review[J]. Composite Structures2018196: 44-54.
[6]
焦李成, 杨淑媛, 刘芳, 等. 神经网络七十年:回顾与展望[J]. 计算机学报201639(8): 1697-1716.
JIAO LichengYANG ShuyuanLIU Fang, et al. Seventy years beyond neural networks: retrospect and prospect[J]. Chinese Journal of Computers201639(8):1697-1716.
[7]
HAKIM S J SRAZAK H A. Structural damage detection of steel bridge girder using artificial neural networks and finite element models[J]. Steel & Composite Structures201314(4): 367-377.
[8]
VAFAEI MALIH S C. Adequacy of first mode shape differences for damage identification of cantilever structures using neural networks[J]. Neural Computing and Applications201830(8): 2509-2518.
[9]
ZHANG X DZHANG Z GLI X F, et al. Damage identification in cable-stayed bridge based on modal analysis and neural networks[C]//AIP Conference Proceedings. Golden, 2007: 1435-1442.
[10]
LECUN YBENGIO YHINTON G. Deep learning[J]. Nature2015521(7553): 436-444.
[11]
DUAN Y FCHEN Q YZHANG H M, et al. CNN-based damage identification method of tied-arch bridge using spatial-spectral information[J]. Smart Structures and Systems201923(5): 507-520.
[12]
PURUNCAJAS BVIDAL YTUTIVÉN C. Vibration-response-only structural health monitoring for offshore wind turbine jacket foundations via convolutional neural networks[J]. Sensors202020(12): 3429.
[13]
邱锡鹏. 神经网络与深度学习[M]. 北京:机械工业出版社, 2020.
QIU Xipeng. Neural Networks and Deep Learning[M]. Beijing: China Machine Press, 2020.
[14]
YU YSI X SHU C H, et al. A review of recurrent neural networks: LSTM cells and network architectures[J]. Neural Computation201931(7): 1235-1270.
[15]
SONY SGAMAGE SSADHU A, et al. Vibration-based multiclass damage detection and localization using long short-term memory networks[J]. Structures202235: 436-451.
[16]
AN Y HOU J P. Experimental and numerical studies on model updating method of damage severity identification utilizing four cost functions[J]. Structural Control and Health Monitoring201320(1): 107-120.
[17]
ZHANG Y FZHU J S. Damage identification for bridge structures based on correlation of the bridge dynamic responses under vehicle load[J]. Structures202133: 68-76.
[18]
闫维明, 顾大鹏, 陈彦江, 等. 基于加速度响应相关性的结构损伤识别方法[J]. 振动与冲击201332(14): 82-86.
YAN WeimingGU DapengCHEN Yanjiang, et al. A method for structural damage detection based on correlation characteristic of acceleration response[J]. Journal of Vibration and Shock201332(14): 82-86.
[19]
杨小森, 闫维明, 陈彦江, 等. 基于振动信号统计特征的损伤识别方法[J]. 公路交通科技201330(12): 99-106.
YANG XiaosenYAN WeimingCHEN Yanjiang, et al. Damage detection method based on statistics characteristics of vibration signal[J]. Journal of Highway and Transportation Research and Development201330(12): 99-106.
[20]
DONAHUE JHENDRICKS L AGUADARRAMA S, et al. Long-term recurrent convolutional networks for visual recognition and description[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Washington, 2015: 2625-2634.
[21]
LEI Y GYANG BJIANG X W, et al. Applications of machine learning to machine fault diagnosis: a review and roadmap[J]. Mechanical Systems and Signal Processing2020138: 106587.
[22]
HOUDT G VMOSQUERA CNÁPOLES G. A review on the long short-term memory model[J]. Artificial Intelligence Review202053(8): 5929-5955.
[23]
YAN W JZHAO M YSUN Q, et al. Transmissibility-based system identification for structural health monitoring: fundamentals, approaches, and applications[J]. Mechanical Systems and Signal Processing2019117: 453-482.
[24]
MERUANE V. Model updating using antiresonant frequencies identified from transmissibility functions[J]. Journal of Sound and Vibration2013332(4): 807-820.
[25]
YAN W JREN W X. Operational modal parameter identification from power spectrum density transmissibility[J]. Computer-Aided Civil and Infrastructure Engineering201227(3):202-217.
[26]
JU H YMA TLI W, et al. Pixelwise asphalt concrete pavement crack detection via deep learning-based semantic segmentation method[J]. Structural Control and Health Monitoring202229(8): e2974.
[27]
伊恩·古德费洛, 约书亚·本吉奥, 亚伦·库维尔. 深度学习[M]. 赵申剑, 黎或君, 符天凡, 等译. 北京:人民邮电出版社, 2021.
[28]
方开泰, 刘民千, 覃红, 等. 均匀试验设计的理论和应用[M]. 北京:科学出版社, 2018.
2024年第37卷第11期
PDF下载
81
40
引用本文
BibTeX
文章信息
doi: 10.16385/j.cnki.issn.1004-4523.2024.11.012
  • 接收时间:2023-04-14
  • 首发时间:2026-02-12
  • 出版时间:2024-11-28
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2023-04-14
  • 修回日期:2023-06-17
基金
国家自然科学基金资助项目(52178276)
福建省自然科学基金资助项目(2021J01601)
福州市科技计划项目(2021-Y-084)
作者信息
    1福州大学土木工程学院,福建 福州 350108
    2福州大学土木工程防震减灾信息化国家地方联合工程研究中心,福建 福州 350108
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/zdgcxb/CN/10.16385/j.cnki.issn.1004-4523.2024.11.012
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏