Article(id=1151591706374844936, tenantId=1146029695717560320, journalId=1146031787341344770, issueId=1151591705854751239, articleNumber=1003-3033(2024)05-0139-08, orderNo=null, doi=10.16265/j.cnki.issn1003-3033.2024.05.0767, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1699891200000, receivedDateStr=2023-11-14, revisedDate=1708185600000, revisedDateStr=2024-02-18, acceptedDate=null, acceptedDateStr=null, onlineDate=1752490458339, onlineDateStr=2025-07-14, pubDate=1716825600000, pubDateStr=2024-05-28, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1752490458339, onlineIssueDateStr=2025-07-14, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1752490458339, creator=13701087609, updateTime=1752490458339, updator=13701087609, issue=Issue{id=1151591705854751239, tenantId=1146029695717560320, journalId=1146031787341344770, year='2024', volume='34', issue='5', pageStart='1', pageEnd='251', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=0, createTime=1752490458214, creator=13701087609, updateTime=1757398693384, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1172178336315985942, tenantId=1146029695717560320, journalId=1146031787341344770, issueId=1151591705854751239, language=EN, specialIssueTitle=, coverIllustrator=, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1172178336315985943, tenantId=1146029695717560320, journalId=1146031787341344770, issueId=1151591705854751239, language=CN, specialIssueTitle=, coverIllustrator=, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=139, endPage=146, ext={EN=ArticleExt(id=1151591706769109513, articleId=1151591706374844936, tenantId=1146029695717560320, journalId=1146031787341344770, language=EN, title=Blade defect detection of U-Net network based on attention mechanism, columnId=1149733269173878863, journalTitle=China Safety Science Journal(CSSJ), columnName=Safety engineering technology, runingTitle=null, highlight=null, articleAbstract=

To solve the issues of wind turbine blades in terms of classification difficulty and blurry segmentation of small defects in surface defect detection,an improved U-Net semantic segmentation network was proposed based on dilated convolution and convolutional attention modules. Based on the encoding-decoding structure of the network model,a transferable VGG16 feature extraction layer was used to replace the encoding part of the U-Net network. Then,a convolutional attention module was added to the skip module between encoding and decoding. The global weight was enhanced by selecting small defect information. Dilated convolution was used in the decoding section to enhance the network's feature extraction ability,and the pre-trained VGG16 model was used to realize transfer learning. The hybrid loss function of Focal and Dice was validated against the models of DeeplabV3+,Pyramid Scene Parsing Network(PSPnet),High-Resolution Network(HRNet),and U-Net. The results showed that the improved U-Net network had higher prediction accuracy in blade defect classification and segmentation tasks,mean intersection over union,mean pixel accuray,and recall values were 83.60%,92.84%,and 88.50%,respectively. The mean intersection over union simulated by the improved U-Net model was 13.98% and 9.38% higher than that by the DeeplabV3+ and standard U-Net model,respectively. Therefore,the proposed model can improve the sensitivity of blade defect detection,effectively reduce false positives of detection results,and provide guidance to wind turbine blade defect detection.

, correspAuthors=Wei LIANG, 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=Lei QI, Ning LI, Wei LIANG, Zheng WANG, Ziliang LIU), CN=ArticleExt(id=1151591739434349563, articleId=1151591706374844936, tenantId=1146029695717560320, journalId=1146031787341344770, language=CN, title=基于注意力机制的U-Net叶片缺陷图像分割, columnId=1149733269727526997, journalTitle=中国安全科学学报, columnName=安全工程技术, runingTitle=null, highlight=null, articleAbstract=

为解决风力发电机叶片表面缺陷检测存在分类困难和微小缺陷分割模糊的难题,构建一种基于扩张卷积和卷积注意力模块的改进U-Net语义分割网络。该网络基于网络模型的编码-解码结构,使用可迁移的VGG16的特征提取层代替U-Net网络的编码部分,在编码-解码之间的跳跃模块加入卷积注意力模块。通过对微小缺陷信息选取加强全局权重,使用扩张卷积增强网络特征,采用VGG16预训练模型实现迁移学习。开展Focal与Dice结合的混合损失函数验证,对比分析DeeplabV3+、PSPnet、HRNet、U-Net这4种模型。结果表明:对于叶片缺陷数据集,改进的U-Net网络模型对叶片缺陷的分类和分割任务具有更高的精度,均交并比、均像素精度和召回率等指标值分别为83.60%、92.84%和88.50%。改进U-Net网络的均交并比值比DeeplabV3+模型高13.98%,比标准U-Net模型高9.38%,能够提高叶片缺陷检测的灵敏度,有效降低检测结果的误报警率,有助于准确检测风机叶片缺陷。

, correspAuthors=梁伟, authorNote=null, correspAuthorsNote=
**梁 伟(1978—),男,陕西绥德人,博士,教授,博士生导师,主要从事油气管道安全检测、站场设备诊断与可靠性评估等方面的研究。E-mail:
, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=19LrOlpwYpQUvMWx12/xNQ==, magXml=MeoTcm8lLTEIoefDhww5Zw==, pdfUrl=null, pdf=rrx7UZ221eCTPMUAayReOA==, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=cgE7fJqyVOm1Yd0MzjtRLw==, mapNumber=null, authorCompany=null, fund=null, authors=

祁 雷 (1986—),男,辽宁葫芦岛人,硕士,工程师,主要从事石油与天然气工程方面的工作。E-mail:

梁伟 教授

, authorsList=祁雷, 李宁, 梁伟, 王峥, 刘子梁)}, authors=[Author(id=1172490954553573441, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=qilei6@cnooc.com.cn, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1172490954620682307, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, authorId=1172490954553573441, language=EN, stringName=Lei QI, firstName=Lei, middleName=null, lastName=QI, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 Clean Energy Branch,CNOOC Energy Development Co.,Ltd.,Tianjin 300459,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1172490954679402564, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, authorId=1172490954553573441, language=CN, stringName=祁雷, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 中海油能源发展股份有限公司 清洁能源分公司 天津 300459, bio={"img":"qL8zbux8jEWTgA+E9aIjbQ==","content":"

祁 雷 (1986—),男,辽宁葫芦岛人,硕士,工程师,主要从事石油与天然气工程方面的工作。E-mail:

"}, bioImg=qL8zbux8jEWTgA+E9aIjbQ==, bioContent=

祁 雷 (1986—),男,辽宁葫芦岛人,硕士,工程师,主要从事石油与天然气工程方面的工作。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1172490954268360759, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, xref=null, ext=[AuthorCompanyExt(id=1172490954276749368, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954268360759, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 Clean Energy Branch,CNOOC Energy Development Co.,Ltd.,Tianjin 300459,China), AuthorCompanyExt(id=1172490954285137977, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954268360759, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 中海油能源发展股份有限公司 清洁能源分公司 天津 300459)])]), Author(id=1172490954746511430, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, 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=1172490954826203208, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, authorId=1172490954746511430, language=EN, stringName=Ning LI, firstName=Ning, middleName=null, lastName=LI, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 Clean Energy Branch,CNOOC Energy Development Co.,Ltd.,Tianjin 300459,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1172490954893312073, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, authorId=1172490954746511430, language=CN, stringName=李宁, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 中海油能源发展股份有限公司 清洁能源分公司 天津 300459, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1172490954268360759, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, xref=null, ext=[AuthorCompanyExt(id=1172490954276749368, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954268360759, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 Clean Energy Branch,CNOOC Energy Development Co.,Ltd.,Tianjin 300459,China), AuthorCompanyExt(id=1172490954285137977, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954268360759, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 中海油能源发展股份有限公司 清洁能源分公司 天津 300459)])]), Author(id=1172490954956226635, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=lw@cup.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1172490955027529806, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, authorId=1172490954956226635, language=EN, stringName=Wei LIANG, firstName=Wei, middleName=null, lastName=LIANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, **, address=2 School of Safety and Ocean Engineering,China University of Petroleum(Beijing),Beijing 102249,China
3 Key Laboratory of Oil and Gas Safety and Emergency Technology,Ministry of Emergency Management,Beijing 102249,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1172490955098832975, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, authorId=1172490954956226635, language=CN, stringName=梁伟, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, **, address=2 中国石油大学(北京) 安全与海洋工程学院,北京 102249
3 应急管理部油气生产安全与应急技术重点实验室,北京 102249, bio={"content":"

梁伟 教授

"}, bioImg=null, bioContent=

梁伟 教授

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1172490954348052538, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, xref=null, ext=[AuthorCompanyExt(id=1172490954356441147, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954348052538, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 School of Safety and Ocean Engineering,China University of Petroleum(Beijing),Beijing 102249,China), AuthorCompanyExt(id=1172490954373218364, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954348052538, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 中国石油大学(北京) 安全与海洋工程学院,北京 102249)]), AuthorCompany(id=1172490954448715837, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, xref=null, ext=[AuthorCompanyExt(id=1172490954457104446, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954448715837, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 Key Laboratory of Oil and Gas Safety and Emergency Technology,Ministry of Emergency Management,Beijing 102249,China), AuthorCompanyExt(id=1172490954461298751, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954448715837, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 应急管理部油气生产安全与应急技术重点实验室,北京 102249)])]), Author(id=1172490955149164625, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, orderNo=3, 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=1172490955220467796, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, authorId=1172490955149164625, language=EN, stringName=Zheng WANG, firstName=Zheng, middleName=null, lastName=WANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2 School of Safety and Ocean Engineering,China University of Petroleum(Beijing),Beijing 102249,China
3 Key Laboratory of Oil and Gas Safety and Emergency Technology,Ministry of Emergency Management,Beijing 102249,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1172490955295965269, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, authorId=1172490955149164625, language=CN, stringName=王峥, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2 中国石油大学(北京) 安全与海洋工程学院,北京 102249
3 应急管理部油气生产安全与应急技术重点实验室,北京 102249, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1172490954348052538, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, xref=null, ext=[AuthorCompanyExt(id=1172490954356441147, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954348052538, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 School of Safety and Ocean Engineering,China University of Petroleum(Beijing),Beijing 102249,China), AuthorCompanyExt(id=1172490954373218364, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954348052538, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 中国石油大学(北京) 安全与海洋工程学院,北京 102249)]), AuthorCompany(id=1172490954448715837, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, xref=null, ext=[AuthorCompanyExt(id=1172490954457104446, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954448715837, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 Key Laboratory of Oil and Gas Safety and Emergency Technology,Ministry of Emergency Management,Beijing 102249,China), AuthorCompanyExt(id=1172490954461298751, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954448715837, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 应急管理部油气生产安全与应急技术重点实验室,北京 102249)])]), Author(id=1172490955350491223, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, orderNo=4, 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=1172490955442765913, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, authorId=1172490955350491223, language=EN, stringName=Ziliang LIU, firstName=Ziliang, middleName=null, lastName=LIU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 Clean Energy Branch,CNOOC Energy Development Co.,Ltd.,Tianjin 300459,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1172490955497291866, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, authorId=1172490955350491223, language=CN, stringName=刘子梁, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 中海油能源发展股份有限公司 清洁能源分公司 天津 300459, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1172490954268360759, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, xref=null, ext=[AuthorCompanyExt(id=1172490954276749368, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954268360759, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 Clean Energy Branch,CNOOC Energy Development Co.,Ltd.,Tianjin 300459,China), AuthorCompanyExt(id=1172490954285137977, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954268360759, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 中海油能源发展股份有限公司 清洁能源分公司 天津 300459)])])], keywords=[Keyword(id=1172490955652481115, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, orderNo=1, keyword=attention mechanism), Keyword(id=1172490955707007068, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, orderNo=2, keyword=U-Net network), Keyword(id=1172490955761533021, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, orderNo=3, keyword=wind turbine blades defect), Keyword(id=1172490955816058974, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, orderNo=4, keyword=image segmentation), Keyword(id=1172490955874779231, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, orderNo=5, keyword=transfer learning), Keyword(id=1172490955929305184, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, orderNo=6, keyword=convolutional block attention module (CBAM)), Keyword(id=1172490955996414049, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, orderNo=1, keyword=注意力机制), Keyword(id=1172490956063522914, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, orderNo=2, keyword=U-Net网络), Keyword(id=1172490956130631779, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, orderNo=3, keyword=风机叶片缺陷), Keyword(id=1172490956193546340, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, orderNo=4, keyword=图像分割), Keyword(id=1172490956264849509, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, orderNo=5, keyword=语义分割), Keyword(id=1172490956403261542, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, orderNo=6, keyword=迁移学习), Keyword(id=1172490956470370407, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, orderNo=7, keyword=卷积块注意力模块(CBAM))], refs=[Reference(id=1172490959402188927, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2022, volume=34, issue=5, pageStart=229, pageEnd=236, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=李志川, 胡鹏, 马佳星, journalName=中国海上油气, refType=null, unstructuredReference=李志川, 胡鹏, 马佳星, 等. 中国海上风电发展现状分析及展望[J]. 中国海上油气, 2022, 34(5):229-236., articleTitle=中国海上风电发展现状分析及展望, refAbstract=null), Reference(id=1172490959494463616, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2022, volume=34, issue=5, pageStart=229, pageEnd=236, url=null, language=null, rfNumber=[1], rfOrder=1, authorNames=LI Zhichuan, HU Peng, MA Jiaxing, journalName=China Offshore Oil and Gas, refType=null, unstructuredReference=LI Zhichuan, HU Peng, MA Jiaxing, et al. Analysis and prospect of offshore wind power development in China[J]. China Offshore Oil and Gas, 2022, 34(5):229-236., articleTitle=Analysis and prospect of offshore wind power development in China, refAbstract=null), Reference(id=1172490959561572481, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2007, volume=10, issue=1, pageStart=1, pageEnd=18, url=null, language=null, rfNumber=[2], rfOrder=2, authorNames=TAVNER P J, XIANG J, SPINATO F, journalName=Wind Energy, refType=null, unstructuredReference=TAVNER P J, XIANG J, SPINATO F. Reliability analysis for wind turbines[J]. Wind Energy, 2007, 10(1):1-18., articleTitle=Reliability analysis for wind turbines, refAbstract=null), Reference(id=1172490959662235778, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2017, volume=11, issue=4, pageStart=411, pageEnd=416, url=null, language=null, rfNumber=[3], rfOrder=3, authorNames=YANG Wenxian, PENG Zhike, WEI Kexiang, journalName=IET Renewable Power Generation, refType=null, unstructuredReference=YANG Wenxian, PENG Zhike, WEI Kexiang, et al. Structural health monitoring of composite wind turbine blades: challenges, issues and potential solutions[J]. IET Renewable Power Generation, 2017, 11(4): 411-416., articleTitle=Structural health monitoring of composite wind turbine blades: challenges, issues and potential solutions, refAbstract=null), Reference(id=1172490959720956035, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[4], rfOrder=4, authorNames=DU Ying, ZHOU Shengxi, JING Xingjian, journalName=Mechanical Systems and Signal Processing, refType=null, unstructuredReference=DU Ying, ZHOU Shengxi, JING Xingjian, et al. Damage detection techniques for wind turbine blades: a review[J]. Mechanical Systems and Signal Processing, 2020,141:DOI:10.1016/j.ymssp.2019.106445., articleTitle=Damage detection techniques for wind turbine blades: a review, refAbstract=null), Reference(id=1172490959792259204, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[5], rfOrder=5, authorNames=HU Wenbo, WANG Weidong, AI Chengbo, journalName=Automation in Construction, refType=null, unstructuredReference=HU Wenbo, WANG Weidong, AI Chengbo, et al. Machine vision-based surface crack analysis for transportation infrastructure[J]. Automation in Construction, 2021,132:DOI: 10.1016/j.autcon.2021.103973., articleTitle=Machine vision-based surface crack analysis for transportation infrastructure, refAbstract=null), Reference(id=1172490959850979461, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2021, volume=55, issue=12, pageStart=119, pageEnd=128, url=null, language=null, rfNumber=[6], rfOrder=6, authorNames=陈法法, 成孟腾, 杨蕴鹏, journalName=西安交通大学学报, refType=null, unstructuredReference=陈法法, 成孟腾, 杨蕴鹏, 等. 融合双注意力机制和U-Net网络的锈蚀图像分割[J]. 西安交通大学学报, 2021, 55(12):119-128., articleTitle=融合双注意力机制和U-Net网络的锈蚀图像分割, refAbstract=null), Reference(id=1172490959918088326, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2021, volume=55, issue=12, pageStart=119, pageEnd=128, url=null, language=null, rfNumber=[6], rfOrder=7, authorNames=CHEN Fafa, CHENG Mengteng, YANG Yunpeng, journalName=Journal of Xi'an Jiaotong University, refType=null, unstructuredReference=CHEN Fafa, CHENG Mengteng, YANG Yunpeng, et al. A segmentation method based on dual attention mechanism and U-Net for corrosion images[J]. Journal of Xi'an Jiaotong University, 2021, 55(12):119-128., articleTitle=A segmentation method based on dual attention mechanism and U-Net for corrosion images, refAbstract=null), Reference(id=1172490960064888967, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2019, volume=11, issue=5, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[7], rfOrder=8, authorNames=XU Donghua, WEN Chuanbo, LIU Jihui, journalName=Renewable Sustainable Energy, refType=null, unstructuredReference=XU Donghua, WEN Chuanbo, LIU Jihui. Wind turbine blade surface inspection based on deep learning and UAV-taken images[J]. Renewable Sustainable Energy, 2019, 11(5):DOI:10.1063/1.5113532., articleTitle=Wind turbine blade surface inspection based on deep learning and UAV-taken images, refAbstract=null), Reference(id=1172490960152969352, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2021, volume=12, issue=8, pageStart=8535, pageEnd=8548, url=null, language=null, rfNumber=[8], rfOrder=9, authorNames=SARKAR D, GUNTURI S K, journalName=Journal of Ambient Intelligence and Humanized Computing, refType=null, unstructuredReference=SARKAR D, GUNTURI S K. Wind turbine blade structural state evaluation by hybrid object detector relying on deep learning models[J]. Journal of Ambient Intelligence and Humanized Computing, 2021, 12(8): 8535-8548., articleTitle=Wind turbine blade structural state evaluation by hybrid object detector relying on deep learning models, refAbstract=null), Reference(id=1172490960224272521, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2021, volume=25, issue=2, pageStart=463, pageEnd=482, url=null, language=null, rfNumber=[9], rfOrder=10, authorNames=MAO Yulin, WANG Shuanxin, YU Dingli, journalName=Intelligent Data Analysis, refType=null, unstructuredReference=MAO Yulin, WANG Shuanxin, YU Dingli, et al. Automatic image detection of multi-type sur face defects on wind turbine blades based on cascade deep learning network[J]. Intelligent Data Analysis, 2021, 25(2):463-482., articleTitle=Automatic image detection of multi-type sur face defects on wind turbine blades based on cascade deep learning network, refAbstract=null), Reference(id=1172490960287187082, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2015, volume=39, issue=4, pageStart=640, pageEnd=651, url=null, language=null, rfNumber=[10], rfOrder=11, authorNames=SHELHAMER E, LONG J, DARRELL T, journalName=IEEE Transactions on Pattern Analysis and Machine Intelligence, refType=null, unstructuredReference=SHELHAMER E, LONG J, DARRELL T. Fully convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 39(4): 640-651., articleTitle=Fully convolutional networks for semantic segmentation, refAbstract=null), Reference(id=1172490960345907339, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2015, volume=null, issue=null, pageStart=234, pageEnd=241, url=null, language=null, rfNumber=[11], rfOrder=12, authorNames=RONNEBERGER O, FISCHER P, BROX T, journalName=International Conference on Medical Image Computing and Computer Assisted Intervention, refType=null, unstructuredReference=RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedical image segmentation[C]. International Conference on Medical Image Computing and Computer Assisted Intervention, 2015:234-241., articleTitle=U-Net: convolutional networks for biomedical image segmentation, refAbstract=null), Reference(id=1172490960408821900, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2017, volume=39, issue=12, pageStart=2481, pageEnd=2495, url=null, language=null, rfNumber=[12], rfOrder=13, authorNames=BADRINARAYANAN V, KENDALL A, CIPOLLA R, journalName=IEEE Transactions on Pattern Analysis and Machine Intelligence, refType=null, unstructuredReference=BADRINARAYANAN V, KENDALL A, CIPOLLA R. SegNet: a deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12):2481-2495., articleTitle=SegNet: a deep convolutional encoder-decoder architecture for image segmentation, refAbstract=null), Reference(id=1172490960471736461, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2016, volume=40, issue=4, pageStart=834, pageEnd=848, url=null, language=null, rfNumber=[13], rfOrder=14, authorNames=CHEN Liangchieh, PAPANDREOU G, KOKKINOS I, journalName=IEEE Transactions on Pattern Analysis and Machine Intelligence, refType=null, unstructuredReference=CHEN Liangchieh, PAPANDREOU G, KOKKINOS I, et al. DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 40(4):834-848., articleTitle=DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs, refAbstract=null), Reference(id=1172490960538845326, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2021, volume=24, issue=9, pageStart=1859, pageEnd=1868, url=null, language=null, rfNumber=[14], rfOrder=15, authorNames=CUI Xiaoning, WANG Qicai, DAI Jinpeng, journalName=Advances in Structural Engineering, refType=null, unstructuredReference=CUI Xiaoning, WANG Qicai, DAI Jinpeng, et al. Intelligent crack detection based on attention mechanism in convolution neural network[J]. Advances in Structural Engineering, 2021, 24(9): 1859-1868., articleTitle=Intelligent crack detection based on attention mechanism in convolution neural network, refAbstract=null), Reference(id=1172490960597565583, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2022, volume=182, issue=null, pageStart=1, pageEnd=16, url=null, language=null, rfNumber=[15], rfOrder=16, authorNames=HACEFENDIOLU K, BASAGA H B, YAVUZ Z, journalName=Renewable Energy, refType=null, unstructuredReference=HACEFENDIOLU K, BASAGA H B, YAVUZ Z, et al. Intelligent ice detection on wind turbine blades using semantic segmentation and class activation map approaches based on deep learning method[J]. Renewable Energy, 2022, 182:1-16., articleTitle=Intelligent ice detection on wind turbine blades using semantic segmentation and class activation map approaches based on deep learning method, refAbstract=null), Reference(id=1172490960647897232, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2023, volume=33, issue=2, pageStart=82, pageEnd=88, url=null, language=null, rfNumber=[16], rfOrder=17, authorNames=夏正洪, 魏汝祥, 李彦冬, journalName=中国安全科学学报, refType=null, unstructuredReference=夏正洪, 魏汝祥, 李彦冬. 改进YOLOv3算法下通航机场场面运动目标检测[J]. 中国安全科学学报, 2023, 33(2):82-88., articleTitle=改进YOLOv3算法下通航机场场面运动目标检测, refAbstract=null), Reference(id=1172490960723394705, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2023, volume=33, issue=2, pageStart=82, pageEnd=88, url=null, language=null, rfNumber=[16], rfOrder=18, authorNames=XIA Zhenghong, WEI Ruxiang, LI Yandong, journalName=China Safety Science Journal, refType=null, unstructuredReference=XIA Zhenghong, WEI Ruxiang, LI Yandong. Moving target detection of general aviation airport based on improved YOLOv3 algorithm[J]. China Safety Science Journal, 2023, 33(2):82-88., articleTitle=Moving target detection of general aviation airport based on improved YOLOv3 algorithm, refAbstract=null), Reference(id=1172490960803086482, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2022, volume=32, issue=5, pageStart=35, pageEnd=40, url=null, language=null, rfNumber=[17], rfOrder=19, authorNames=何静, 侯娜, 张昌凡, journalName=中国安全科学学报, refType=null, unstructuredReference=何静, 侯娜, 张昌凡, 等. 基于金字塔拆分注意力的列车轮对踏面损伤诊断[J]. 中国安全科学学报, 2022, 32(5):35-40., articleTitle=基于金字塔拆分注意力的列车轮对踏面损伤诊断, refAbstract=null), Reference(id=1172490960878583955, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2022, volume=32, issue=5, pageStart=35, pageEnd=40, url=null, language=null, rfNumber=[17], rfOrder=20, authorNames=HE Jing, HOU Na, ZHANG Changfan, journalName=China Safety Science Journal, refType=null, unstructuredReference=HE Jing, HOU Na, ZHANG Changfan, et al. Diagnosis of train wheelset tread damage based on EPSA-ResNet[J]. China Safety Science Journal, 2022, 32(5):35-40., articleTitle=Diagnosis of train wheelset tread damage based on EPSA-ResNet, refAbstract=null), Reference(id=1172490960945692820, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=3, pageEnd=19, url=null, language=null, rfNumber=[18], rfOrder=21, authorNames=WOO S, PARK J, LEE J Y, journalName=Proceedings of the European Conference on Computer Vision (ECCV), refType=null, unstructuredReference=WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]. Proceedings of the European Conference on Computer Vision (ECCV), 2018: 3-19., articleTitle=CBAM: convolutional block attention module, refAbstract=null), Reference(id=1172490961008607381, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2022, volume=59, issue=2, pageStart=1, pageEnd=17, url=null, language=null, rfNumber=[19], rfOrder=22, authorNames=张欢, 仇大伟, 冯毅博, journalName=激光与光电子学进展, refType=null, unstructuredReference=张欢, 仇大伟, 冯毅博, 等. U-Net模型改进及其在医学图像分割上的研究综述[J]. 激光与光电子学进展, 2022, 59(2):1-17., articleTitle=U-Net模型改进及其在医学图像分割上的研究综述, refAbstract=null), Reference(id=1172490961071521942, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, doi=null, pmid=null, pmcid=null, year=2022, volume=59, issue=2, pageStart=1, pageEnd=17, url=null, language=null, rfNumber=[19], rfOrder=23, authorNames=ZHANG Huan, QIU Dawei, FENG Yibo, journalName=Laser & Optoelectronics Progress, refType=null, unstructuredReference=ZHANG Huan, QIU Dawei, FENG Yibo, et al. Improved U-Net models and its applications in medical image segmentation: a review[J]. Laser & Optoelectronics Progress, 2022, 59(2):1-17., articleTitle=Improved U-Net models and its applications in medical image segmentation: a review, refAbstract=null)], funds=[Fund(id=1172490959226028158, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, awardId=GD2021ZCAF0021, language=CN, fundingSource=中海油重大科技项目(GD2021ZCAF0021), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1172490954268360759, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, xref=null, ext=[AuthorCompanyExt(id=1172490954276749368, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954268360759, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 Clean Energy Branch,CNOOC Energy Development Co.,Ltd.,Tianjin 300459,China), AuthorCompanyExt(id=1172490954285137977, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954268360759, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 中海油能源发展股份有限公司 清洁能源分公司 天津 300459)]), AuthorCompany(id=1172490954348052538, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, xref=null, ext=[AuthorCompanyExt(id=1172490954356441147, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954348052538, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 School of Safety and Ocean Engineering,China University of Petroleum(Beijing),Beijing 102249,China), AuthorCompanyExt(id=1172490954373218364, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954348052538, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 中国石油大学(北京) 安全与海洋工程学院,北京 102249)]), AuthorCompany(id=1172490954448715837, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, xref=null, ext=[AuthorCompanyExt(id=1172490954457104446, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954448715837, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 Key Laboratory of Oil and Gas Safety and Emergency Technology,Ministry of Emergency Management,Beijing 102249,China), AuthorCompanyExt(id=1172490954461298751, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, companyId=1172490954448715837, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 应急管理部油气生产安全与应急技术重点实验室,北京 102249)])], figs=[ArticleFig(id=1172490956873023592, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, label=Fig.1, caption=U-Net network structure, figureFileSmall=8MILe26sOcHDNj4nm1roCA==, figureFileBig=QnDSOcxWWGTx3qNDJ/n2OA==, tableContent=null), ArticleFig(id=1172490956977881193, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, label=图1, caption=U-Net的网络结构, figureFileSmall=8MILe26sOcHDNj4nm1roCA==, figureFileBig=QnDSOcxWWGTx3qNDJ/n2OA==, tableContent=null), ArticleFig(id=1172490957116293226, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, label=Fig.2, caption=CBAM attention module, figureFileSmall=W1P7eVLSW8xnQ4VKqObjUQ==, figureFileBig=lfxA8BC72Lbbolyxg07xdA==, tableContent=null), ArticleFig(id=1172490957233733739, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, label=图2, caption=CBAM注意力机制, figureFileSmall=W1P7eVLSW8xnQ4VKqObjUQ==, figureFileBig=lfxA8BC72Lbbolyxg07xdA==, tableContent=null), ArticleFig(id=1172490957313425516, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, label=Fig.3, caption=Channel attention module, figureFileSmall=zqQNdDjSAVlDEn6rl8BcIw==, figureFileBig=KdnRxgYn4wPNC3dLxI0tSw==, tableContent=null), ArticleFig(id=1172490957393117293, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, label=图3, caption=通道注意力模块, figureFileSmall=zqQNdDjSAVlDEn6rl8BcIw==, figureFileBig=KdnRxgYn4wPNC3dLxI0tSw==, tableContent=null), ArticleFig(id=1172490957464420462, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, label=Fig.4, caption=Spatial attention module, figureFileSmall=JceyqZ/u92EsU1TSDhGA6g==, figureFileBig=acCV7Yr85LIY61I0v2ut4Q==, tableContent=null), ArticleFig(id=1172490957531529327, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, label=图4, caption=空间注意力模块, figureFileSmall=JceyqZ/u92EsU1TSDhGA6g==, figureFileBig=acCV7Yr85LIY61I0v2ut4Q==, tableContent=null), ArticleFig(id=1172490957640581232, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, label=Fig.5, caption=Improved U-Net network structure, figureFileSmall=OBiYXd/0Mr8xPM6BRmn3Vw==, figureFileBig=HJBjNvqRN+7EkOiLueAYbQ==, tableContent=null), ArticleFig(id=1172490957707690097, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, label=图5, caption=改进U-Net网络结构, figureFileSmall=OBiYXd/0Mr8xPM6BRmn3Vw==, figureFileBig=HJBjNvqRN+7EkOiLueAYbQ==, tableContent=null), ArticleFig(id=1172490957778993266, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, label=Fig.6, caption=Training and testing process of semantic segmentation model, figureFileSmall=JZnn+/wlpHncDvoxwdX3xw==, figureFileBig=u8r/5VaCnHY6rn355xJNlA==, tableContent=null), ArticleFig(id=1172490957858685043, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, label=图6, caption=语义分割模型训练和测试过程, figureFileSmall=JZnn+/wlpHncDvoxwdX3xw==, figureFileBig=u8r/5VaCnHY6rn355xJNlA==, tableContent=null), ArticleFig(id=1172490957917405300, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, label=Fig.7, caption=Image enhancement results, figureFileSmall=5XMUXVRk2uEBkCX0gQlUEw==, figureFileBig=OibdHi0doqdi2xmN2Xptbw==, tableContent=null), ArticleFig(id=1172490958026457205, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, label=图7, caption=图像增强结果, figureFileSmall=5XMUXVRk2uEBkCX0gQlUEw==, figureFileBig=OibdHi0doqdi2xmN2Xptbw==, tableContent=null), ArticleFig(id=1172490958101954678, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, label=Fig.8, caption=Performance comparison of different defect image segmentation methods, figureFileSmall=hNC/Kz3MPLpuldshkW0lhg==, figureFileBig=Bjekn9HedLeVOqftNo0HPg==, tableContent=null), ArticleFig(id=1172490958194229367, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, label=图8, caption=不同缺陷分割方法的效果对比, figureFileSmall=hNC/Kz3MPLpuldshkW0lhg==, figureFileBig=Bjekn9HedLeVOqftNo0HPg==, tableContent=null), ArticleFig(id=1172490958273921144, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, label=Fig.9, caption=Comparison of model segmentation performance, figureFileSmall=vHw7x3ENBsOlQOGUiaaKTQ==, figureFileBig=4TwvnPkGUs4k/VcCPDkXdQ==, tableContent=null), ArticleFig(id=1172490958382973049, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, label=图9, caption=模型分割性能对比, figureFileSmall=vHw7x3ENBsOlQOGUiaaKTQ==, figureFileBig=4TwvnPkGUs4k/VcCPDkXdQ==, tableContent=null), ArticleFig(id=1172490958584299642, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, label=Table 1, caption=

Simulated intersection over union indicators of each model for different defects

, figureFileSmall=null, figureFileBig=null, tableContent=
分割模型 缺陷类别 均交并比
背景 裂纹 剥落 纤维断裂 腐蚀
DeeplabV3+ 98.69 30.55 72.72 86.23 59.89 69.62
HRNet 98.88 59.32 83.84 91.31 70.86 80.84
PSPnet 98.63 42.87 71.11 88.55 45.71 69.37
U-Net 98.70 40.87 73.13 84.67 73.75 74.22
文中模型 99.25 66.58 87.00 91.50 73.66 83.60
), ArticleFig(id=1172490958714323067, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, label=表1, caption=

各模型对不同缺陷的交并比指标计算结果

, figureFileSmall=null, figureFileBig=null, tableContent=
分割模型 缺陷类别 均交并比
背景 裂纹 剥落 纤维断裂 腐蚀
DeeplabV3+ 98.69 30.55 72.72 86.23 59.89 69.62
HRNet 98.88 59.32 83.84 91.31 70.86 80.84
PSPnet 98.63 42.87 71.11 88.55 45.71 69.37
U-Net 98.70 40.87 73.13 84.67 73.75 74.22
文中模型 99.25 66.58 87.00 91.50 73.66 83.60
), ArticleFig(id=1172490958907261052, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=EN, label=Table 2, caption=

Comparison of defect image segmentation performance of different semantic segmentation models

, figureFileSmall=null, figureFileBig=null, tableContent=
分割模型 准确度 召回率 均交并比 均像素精度
DeeplabV3+ 95.63 79.65 69.62 80.02
HRNet 99.16 86.60 80.70 90.85
PSPnet 98.65 73.71 69.37 89.58
U-Net 98.62 80.64 74.22 87.27
文中模型 99.26 88.50 83.60 92.84
), ArticleFig(id=1172490958995341437, tenantId=1146029695717560320, journalId=1146031787341344770, articleId=1151591706374844936, language=CN, label=表2, caption=

不同语义分割模型的缺陷图像分割性能对比

, figureFileSmall=null, figureFileBig=null, tableContent=
分割模型 准确度 召回率 均交并比 均像素精度
DeeplabV3+ 95.63 79.65 69.62 80.02
HRNet 99.16 86.60 80.70 90.85
PSPnet 98.65 73.71 69.37 89.58
U-Net 98.62 80.64 74.22 87.27
文中模型 99.26 88.50 83.60 92.84
)], attaches=null, journal=Journal(id=1123942128916217864, delFlag=0, nameCn=中国安全科学学报, nameEn=China Safety Science Journal, nameHistory1=null, nameHistory2=null, issn=1003-3033, eissn=, cn=11-2865/X, coden=null, periodic=0, language=CN, oaType=0, 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=fkqsFM6VKlHC4gCtS5XqTw==, journalPrice=null, startedYear=null, abbrevIsoEn=Chin Saf Sci J, journalRemark=null, publicationField=null, createdTime=null, updatedTime=1754269350027, createdBy=null, updatedBy=13701087609, firstLetterCn=C, firstLetterEn=C, subjectCode=Engineering, subjectName=工程, subjectCodeEn=Engineering, subjectNameEn=null, picCn=fkqsFM6VKlHC4gCtS5XqTw==, picEn=SHn9HgqSxtJrOcAxqD++4Q==, jcr=null, cjcr=null, exts=[JournalExt(id=1159052918994595848, language=CN, name=中国安全科学学报, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.cssjj.com.cn/, createdTime=1754269350050, updatedTime=1754269350050, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=http://www.cssjj.com.cn/CN/column/item15.shtml, submissionAuthorUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionReviewUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionCeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionAeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, option={"copyright":""}), JournalExt(id=1159052919040733193, language=EN, name=China Safety Science Journal, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.cssjj.com.cn/EN/1003-3033/home.shtml, createdTime=1754269350061, updatedTime=1754269350061, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=https://synbioj.cip.com.cn/EN/column/column3.shtml, submissionAuthorUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionReviewUrl=https://zgaqkxxbauthor.manuscriptcloud.com/login, submissionCeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, submissionAeEditorUrl=https://zgaqkxxbeditor.manuscriptcloud.com/login, option={"copyright":""})], databaseList=null, tenantJournalId=1146031787341344770, websiteList=[Website(id=1148243202345263519, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146031787341344770, 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/zgaqkxxb/CN, language=CN, createTime=1751692112766, createBy=18614031015, updateTime=1753502583634, updateBy=18614031015, name=《中国安全科学学报》中文站点, tplId=1146099689490845704, title=中国安全科学学报, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1148618794941046792, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=articleTextType, value=kx, createTime=1751781661020, updateTime=1751781661020, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794911686661, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=banner, value=null, createTime=1751781661012, updateTime=1751781661012, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794894909444, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=tui0IVO9FMwB61HHtX5scg==, createTime=1751781661008, updateTime=1751781661008, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794932658183, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1751781661017, updateTime=1751781661017, creator=18614031015, updator=18614031015), WebsiteProps(id=1148618794924269574, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202345263519, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1751781661015, updateTime=1751781661015, creator=18614031015, updator=18614031015)]), Website(id=1155836763751993353, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146031787341344770, 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/zgaqkxxb/EN, language=EN, createTime=1753502558893, createBy=18614031015, updateTime=1753524450387, updateBy=18614031015, name=《中国安全科学学报》英文站点, tplId=1146101810881728533, title=China Safety Science Journal, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1155895925743669425, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=articleTextType, value=kx, createTime=1753516664205, updateTime=1753516664205, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925722697902, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=banner, value=null, createTime=1753516664200, updateTime=1753516664200, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925714309293, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=tui0IVO9FMwB61HHtX5scg==, createTime=1753516664198, updateTime=1753516664198, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925735280816, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1753516664203, updateTime=1753516664203, creator=18614031015, updator=18614031015), WebsiteProps(id=1155895925731086511, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155836763751993353, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1753516664202, updateTime=1753516664202, creator=18614031015, updator=18614031015)])], journalTitle=中国安全科学学报, weixinUrl=null, journalUrl=null, iacademicId=null, status=0, seqNo=null, journalTitleEn=China Safety Science Journal, journalPhotoCn=fkqsFM6VKlHC4gCtS5XqTw==, journalPhotoEn=SHn9HgqSxtJrOcAxqD++4Q==, journalFirstLetter=C, journalRecommend=null, journalNew=null, journalCollection=1, 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=null, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/CN/10.16265/j.cnki.issn1003-3033.2024.05.0767, detailUrlEn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/EN/10.16265/j.cnki.issn1003-3033.2024.05.0767, pdfUrlCn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/CN/PDF/10.16265/j.cnki.issn1003-3033.2024.05.0767, pdfUrlEn=https://castjournals.cast.org.cn/joweb/zgaqkxxb/EN/PDF/10.16265/j.cnki.issn1003-3033.2024.05.0767, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
基于注意力机制的U-Net叶片缺陷图像分割
收藏切换
PDF下载
祁雷 1 , 李宁 1 , 梁伟 2, 3, ** , 王峥 2, 3 , 刘子梁 1
中国安全科学学报 | 安全工程技术 2024,34(5): 139-146
收起
收藏切换
中国安全科学学报 | 安全工程技术 2024, 34(5): 139-146
基于注意力机制的U-Net叶片缺陷图像分割
全屏
祁雷1 , 李宁1, 梁伟2, 3, ** , 王峥2, 3, 刘子梁1
作者信息
  • 1 中海油能源发展股份有限公司 清洁能源分公司 天津 300459
  • 2 中国石油大学(北京) 安全与海洋工程学院,北京 102249
  • 3 应急管理部油气生产安全与应急技术重点实验室,北京 102249
  • 祁 雷 (1986—),男,辽宁葫芦岛人,硕士,工程师,主要从事石油与天然气工程方面的工作。E-mail:

    梁伟 教授

通讯作者:

**梁 伟(1978—),男,陕西绥德人,博士,教授,博士生导师,主要从事油气管道安全检测、站场设备诊断与可靠性评估等方面的研究。E-mail:
Blade defect detection of U-Net network based on attention mechanism
Lei QI1 , Ning LI1, Wei LIANG2, 3, ** , Zheng WANG2, 3, Ziliang LIU1
Affiliations
  • 1 Clean Energy Branch,CNOOC Energy Development Co.,Ltd.,Tianjin 300459,China
  • 2 School of Safety and Ocean Engineering,China University of Petroleum(Beijing),Beijing 102249,China
  • 3 Key Laboratory of Oil and Gas Safety and Emergency Technology,Ministry of Emergency Management,Beijing 102249,China
出版时间: 2024-05-28 doi: 10.16265/j.cnki.issn1003-3033.2024.05.0767
文章导航
收藏切换

为解决风力发电机叶片表面缺陷检测存在分类困难和微小缺陷分割模糊的难题,构建一种基于扩张卷积和卷积注意力模块的改进U-Net语义分割网络。该网络基于网络模型的编码-解码结构,使用可迁移的VGG16的特征提取层代替U-Net网络的编码部分,在编码-解码之间的跳跃模块加入卷积注意力模块。通过对微小缺陷信息选取加强全局权重,使用扩张卷积增强网络特征,采用VGG16预训练模型实现迁移学习。开展Focal与Dice结合的混合损失函数验证,对比分析DeeplabV3+、PSPnet、HRNet、U-Net这4种模型。结果表明:对于叶片缺陷数据集,改进的U-Net网络模型对叶片缺陷的分类和分割任务具有更高的精度,均交并比、均像素精度和召回率等指标值分别为83.60%、92.84%和88.50%。改进U-Net网络的均交并比值比DeeplabV3+模型高13.98%,比标准U-Net模型高9.38%,能够提高叶片缺陷检测的灵敏度,有效降低检测结果的误报警率,有助于准确检测风机叶片缺陷。

注意力机制  /  U-Net网络  /  风机叶片缺陷  /  图像分割  /  语义分割  /  迁移学习  /  卷积块注意力模块(CBAM)

To solve the issues of wind turbine blades in terms of classification difficulty and blurry segmentation of small defects in surface defect detection,an improved U-Net semantic segmentation network was proposed based on dilated convolution and convolutional attention modules. Based on the encoding-decoding structure of the network model,a transferable VGG16 feature extraction layer was used to replace the encoding part of the U-Net network. Then,a convolutional attention module was added to the skip module between encoding and decoding. The global weight was enhanced by selecting small defect information. Dilated convolution was used in the decoding section to enhance the network's feature extraction ability,and the pre-trained VGG16 model was used to realize transfer learning. The hybrid loss function of Focal and Dice was validated against the models of DeeplabV3+,Pyramid Scene Parsing Network(PSPnet),High-Resolution Network(HRNet),and U-Net. The results showed that the improved U-Net network had higher prediction accuracy in blade defect classification and segmentation tasks,mean intersection over union,mean pixel accuray,and recall values were 83.60%,92.84%,and 88.50%,respectively. The mean intersection over union simulated by the improved U-Net model was 13.98% and 9.38% higher than that by the DeeplabV3+ and standard U-Net model,respectively. Therefore,the proposed model can improve the sensitivity of blade defect detection,effectively reduce false positives of detection results,and provide guidance to wind turbine blade defect detection.

attention mechanism  /  U-Net network  /  wind turbine blades defect  /  image segmentation  /  transfer learning  /  convolutional block attention module (CBAM)
祁雷, 李宁, 梁伟, 王峥, 刘子梁. 基于注意力机制的U-Net叶片缺陷图像分割. 中国安全科学学报, 2024 , 34 (5) : 139 -146 . DOI: 10.16265/j.cnki.issn1003-3033.2024.05.0767
Lei QI, Ning LI, Wei LIANG, Zheng WANG, Ziliang LIU. Blade defect detection of U-Net network based on attention mechanism[J]. China Safety Science Journal(CSSJ), 2024 , 34 (5) : 139 -146 . DOI: 10.16265/j.cnki.issn1003-3033.2024.05.0767
风能作为一种无污染、可再生的新型清洁能源,具有广阔的开发前景和商业价值[1]。叶片是风力发电机中捕获风能的关键部件,也是风机的易损部件之一,据统计,叶片损伤故障占所有风机部件故障的7%[2]。如果叶片在运行中出现结构缺陷并随时间蔓延,将导致结构失效,造成维修或更换叶片的经济损失及非计划停机对发电效率的影响[3]。早期检测与诊断风电机组叶片腐蚀、缺陷和裂纹等是本领域研究的热点与难点。
风电机组在运行中不仅受复杂载荷作用,而且受湿度、雷击、覆冰等环境影响[4]。风机叶片上的主要缺陷类型有表面裂纹、沿纤维开裂、粘合剂脱粘(材料失去相互粘附能力)和表面腐蚀等。目前,对风力发电机叶片的检测方法主要有人工检测、传感器监测和基于机器视觉等,人工检测可能发生意外事故造成健康风险,传感器监测容易脱落、采集数据不稳定。随着计算机技术的创新与发展,基于数据驱动的机器视觉技术在结构健康监测领域得到了广泛应用。
基于机器视觉的缺陷检测方法主要分为图像处理、机器学习和深度学习等3类[5]。图像处理方法能够处理浅层次的图像特征,机器学习可以感知的特征掌握识别规则,深度学习方法实现了自动提取特征,获取丰富且深入的特征,能够更全面、深入地分析种类繁多的缺陷[6]。在风力发电机缺陷检测领域,许多研究者已将深度学习方法应用于风力发电机叶片和塔架的表面缺陷检测。XU Donghua等[7]使用风机叶片缺陷数据集训练了VGG模型,利用乘数交替法压缩模型,实现风机叶片缺陷检测与分类。SARKAR等[8]提出一种超分辨率卷积神经网络(Super Resolution Convolutional Neural Networks, SRCNN)方法,将其与YOLO-v3网络结合,实现高精度检测风机叶片缺陷。MAO Yulin等[9]提出一种改进基于区域的卷积神经网络(Region-based Convolutional Neural Networks, R-CNN)模型,利用迁移学习、优化卷积块等方法减少网络训练过程中的信息丢失,提高了缺陷检测精度。随着卷积神经网络(Convolutional Neural Networks, CNN)的发展,利用像素级检测算法实现精细化检测也成为结构健康监测领域的发展热点。
SHELHAMER等[10]提出用于图像语义分割的全卷积神经网络(Fully Convolutional Networks, FCN)。FCN从抽象的特征中判断每个像素的类别,实现精确的图像分割。目前,许多语义分割网络都是基于FCN,如U-Net网络[11]和Segnet网络[12]采用基于FCN的编码-解码结构,Deeplab网络[13]采用空洞卷积结构,其中,U-Net网络应用最为广泛。CUI Xiaoning等[14]结合注意力模块与U-Net网络模型,利用注意力机制关注关键区域,提高了混凝土裂纹的检测准确度,HACEFENDIOLU等[15]利用开源数据集Imagenet预训练多个CNN模型,与基于通道注意力机制的可视化技术(Score-Channel Attention Module, Score-CAM)结合,实现风机叶片结冰区域的定位。综上,目前的大型风力发电机组缺陷检测中,在局部缺陷的检测中效果较好,可以较好地识别出缺陷类型或缺陷位置。但尚缺乏反映结构的整体损伤情况,且由于不同缺陷类型之间的尺寸相差过大,导致的裂纹等微小缺陷准确检测方法。
鉴于此,笔者拟基于VGG16网络模型迁移,提升图像分割任务的效率和准确度,采用卷积块注意力模块(Convolutional Block Attention Module, CBAM)改进U-Net网络的跳跃连接部分,减少与缺陷区域无关的特征干扰,使用扩张卷积加强特征提取能力,进一步增强U-Net的稳定性和泛化能力,在复合缺陷类型的数据集上验证模型准确率,以期实现复杂缺陷和微小缺陷的准确分类与精确分割。
U-Net网络于2015年被提出,已证明是一种有效的语义分割方法,用于图像中目标区域的像素级分割。
其编码器有4个编码块,编码块由2个卷积层和1个最大池化层组成,每个编码块对应一次下采样。原始图像每进行一次下采样,特征通道数翻倍,特征图尺寸减半。
W-解码器主要通过卷积与反卷积将特征映射至分割图像中,原始U-Net网络的解码器有4个解码块,解码块由2个卷积层和1个反卷积层组成,每个解码块对应一次上采样。特征图进行一次上采样,特征通道数减半,特征图尺寸翻倍。编码器与解码器之间有跳跃连接,最终由网络末端的预测部分对所有像素完成分类。
数据集和标签的质量是影响深度学习网络性能的重要因素之一。目前还未有大规模的风力发电机叶片表面缺陷数据集。在目标领域数据量不足的情况下,采用迁移学习可节省数据需求和计算资源,减少训练成本和时间[16]
利用预训练VGG16网络模型迁移学习,可提高U-Net在风机叶片缺陷检测方面的性能,改进U-Net网络的编码器网络部分,使用ImageNet数据集对其预训练,优化U-Net编码器网络模块。
CBAM模块是一种应用于深度学习中的注意力机制,主要用于CNN模型中改进网络的感受野,提升网络的识别性能。2018年,WOO等[17]提出在ResNet和MobieNet等网络中添加CBAM注意力模块,加强了目标的识别效果。
CBAM的网络结构包含通道注意力机制(Channel Attention Module, CAM)和空间注意力机制(Spatial Attention Module, SAM)2个部分。CBAM能够自适应分配特征权值,关注图像中的重要区域。因此,CBAM模块适用于语义分割等机器视觉检测任务。CBAM注意力机制如图2所示。
注意力机制对神经网络中不同通道特征施加,由全局最大池化层、全局平均池化层、全连接层和sigmoid函数组成。其工作原理为:输入1个H×W×C特征图像,分别经过最大池化和平均池化层后,得到2个1×1×C的特征图,通过共享全连接层得到各通道的权重矢量,权值相加后经过sigmoid激活函数后生成通道注意力特征。
将通道注意力特征与最初输入相乘得到H×W×C特征图,通道注意力机制能够加权不同通道的特征,使模型能够利用不同通道的信息[18],通道注意力模块如图3所示。
利用空间注意力机制对网络中不同空间位置的特征迁移学习,通过全局最大池化层、全局平均池化层、卷积层和sigmoid函数实现。其工作原理为:输入一个H×W×C的特征图像,经过最大池化和平均池化层后,得到2个H×W×1的特征图。将2张特征图通道拼接后,经过1次卷积,得到1个代表空间注意力图的H×W×1特征图,经过sigmoid激活函数后生成空间注意力特征,将空间注意力特征与最初输入相乘得到H×W×C特征图,空间注意力机制能够使模型集中关注重要空间位置的特征,如图4所示。
扩张卷积是一种针对下采样缩小图像分辨率问题提出的一种卷积思路。在原本卷积核中插入空隙,通过扩张率参数控制卷积核中空隙大小,增加卷积核的视野域,使网络在原有的参数量下能够更快收敛[19]。利用扩张卷积替代原本卷积块,可更好地提取深层特征,提升模型的图像分割效果。
基于传统U-Net语义分割网络,设计一种用于风机叶片缺陷检测与分类的语义分割模型。改进后的U-Net算法框架如图5所示。
网络采用编码器-解码器结构,分别对应特征图下采样、上采样部分。网络的左半部分是编码器网络,右半部分是解码器网络,编码器与解码器之间使用跳跃层连接全局和局部特征。
采用VGG16网络的前13次卷积,从输入图像中提取特征。输入图像经过5次卷积、4次下采样,实现特征浓缩。其中,第1、2次卷积操作使用2组3×3卷积、批归一化、激活层提取特征。
第3、4、5次卷积操作使用3组3×3卷积,加强特征提取能力。训练过程中,冻结预训练模型的权值,目的是优先对加强特征提取网络的参数训练,以加强训练效率。
对解码器进行上采样,对双线性插值进行尺寸扩充,与来自下采样的特征图共同输入注意力机制,将输出的加强信息特征图输入至扩张卷积模块中进一步提取深层语义特征,再反卷积操作,经过Sigmoid激活函数后生成预测概率图。
在缺陷图像的分割中,深度学习和机器视觉检测技术有更高的稳定性,准确的语义分割可以识别像素级目标图像,模型训练和验证过程,如图6所示。
利用Dice损失函数衡量预测图像与分割图像之间的相似度,计算预测图像与标签图像中重叠像素点的比例,评估预测准确性。
Dice损失函数LD的数学表达式如下:
L D = 1 - 2 i = 1 N g i p i i = 1 N g i + i = 1 N p i
式中: g i为像素i的真实类别; p i为模型对像素i的预测类别;N为图像的总像素点数。
Focal损失可避免分类样本损失函数过大。Focal损失函数LF的数学表达式如下:
L F = - α ( 1 - p i ) γ l n p i g i = 1 - ( 1 - α ) p i γ l n ( 1 - p i ) g i = 0
式中 α γ为可调节平衡因子。
为衡量语义分割模型的性能和准确度,采用准确度 A C、召回率 R E、交并比I、均交并比 M I、像素准确度X、均像素精度 M X等评估指标评估模的缺陷分类和图像分割能力。计算公式如下:
A C = T p T p + F p
R E = T p T p + F n
I = T p T p + F p + F n
M I = i = 1 m I m
X = T p + T n T p + F p + T n + F n
M X = i = 1 m X m
式中: T p为真阳性像素数目; F p为假阳性像素数目; T n为真阴性像素数目; F n为假阴性像素数目;m为类别数目。
数据集来源于风机叶片表面缺陷数据集,截取123张风力发电机叶片的高分辨率图像,图像尺寸为8 688×5 792。
数据集包含裂纹、腐蚀、剥落和纤维断裂等缺陷类型多种表面结构缺陷,其中,裂纹和腐蚀的缺陷尺寸相对其他缺陷较小,将其归类于微小缺陷。裂纹较细的横纵向裂口,腐蚀在叶片表面呈现黑色或黄褐色斑点,剥落为不影响内部结构的表面划痕、磨损、凹陷或凸起,纤维断裂是层压板内部分离纤维的开裂或断裂。
标注图像中4种缺陷类型,裂纹、腐蚀、剥落和纤维断裂损伤等缺陷分别由红、蓝、绿、黄色填充,按照8∶1∶1的比例将数据集划分为训练集(99张)、验证集(12张)、测试集(12张)。
为增加数据的数量和多样性,随机旋转、平移、缩放、翻转、亮度调节、加噪训练集和验证集中的每个样本。
每个样本进行3次数据增强,生成随机、多样的样本,以提高语义分割模型的识别和泛化能力。数据增强后,训练集图像扩充至297张,验证集图像扩充至36张。对测试集不采取数据增强,直接用于模型的评估,如图7所示。
使用ImageNet数据集训练VGG16网络,得到预训练权重,迭代次数为100轮,批设置为4。采用Adam优化算法,cos学习率下降方式,初始学习率0.000 1。
选取经典的语义分割算法(如DeeplabV3+,Pyramid Scene Parsing Network(PSPnet), High-Resolution Network(HRNet),U-Net)对照测试,比较不同深度学习模型对缺陷的分割效果,分割测试图像的缺陷,如图8所示。
图8中,每一列都是不同种类缺陷原图与不同模型的分割结果。通过对比分割结果可以发现,DeeplabV3+对剥落边缘和裂纹的识别效果有限,在复合缺陷分割结果中对于边缘区域分割性能有限,PSPnet能够实现缺陷的正确分类,但只能够粗略分割出缺陷区域。
结果表明:HRNet和U-Net能正确分类缺陷,PSPnet能更精细地分割除裂纹的其他缺陷区域,HRNet和U-Net能正确分类不同种类的裂纹缺陷。
交并比和均交并比是语义分割任务中常用的评估指标。计算以上语义分割网络对背景及不同种类缺陷的交并比及均交并比值,结果见表1
对比可知:背景、纤维断裂与剥落缺陷的I值较高,这是因为这3类占图像像素比重较大,特征更明显,相对于其他类更容易区分,腐蚀与裂纹2类缺陷占图像像素的比重小,在模型训练过程中很难区分,因此,这2类缺陷的I值偏低。
在语义分割网络中,DeeplabV3+和PSPnet对所有缺陷类型的分割性能均有限,U-Net网络对裂纹的分割效果远低于平均值,HRNet对裂纹和腐蚀等微小缺陷的分割效果有限,文中模型在5种类别上分割效果均优于其他网络模型。采用2.4节中的评价指标定量评估不同语义分割模型,各算法的评价指标计算结果见表2
试验比对可知:文中模型在各指标上明显优于传统语义分割模型,在准确度上也有优势。
为验证模型结构的有效性及可行性,在语义分割任务中,通过消融试验,分析模型中不同部分对模型性能的贡献。
分别对传统U-Net、U-Net+VGG16迁移学习,U-Net+迁移学习+注意力机制+扩张卷积网络模型上进行缺陷分割,记录每个模型的性能表现。根据召回率、均交并比、均像素精度等指标评估模型性能,结果如图9所示。
对比可知:传统U-Net模型的分割准确率在3种模型中最低,添加VGG16模型迁移学习后,召回率、均交并比与均像素精度等指标分别提高3.31%、5.03%和 4.62%,这表明添加迁移学习对风机缺陷小样本数据集的有效性。
文中模型在召回率、均交并比与像素精度等指标上分别提高7.86%、9.38%和5.57%。同时,对比添加迁移学习的U-Net模型,各指标的值均有提升。表明基于U-Net语义分割模型用于叶片缺陷检测、分类与分割的有效性。
1) 基于U-Net网络结构,采用VGG16迁移学习策略改进网络编码器部分,利用扩张卷积块代替解码器中的卷积提取深层特征,在网络跳跃连接部分增加CBAM注意力机制促进特征融合。将改进U-Net语义分割模型与不同分割算法对比,改进U-Net语义分割模型对同一组缺陷图像的分割效果优势明显,缺陷识别准确,分割边界清晰。
2) 从不同算法分割效果和评估指标验证对比,改进U-Net语义分割模型的准确度达到99.26%,召回率达到88.50%,均交并比达到83.60%,均像素精度达到92.84%,相比于其他分割算法,改进U-Net语义分割模型在叶片缺陷分类与分割任务中性能最优。
3) 依据消融试验,U-Net+迁移学习策略的均交并比值比标准U-Net模型提高5.03%,表明采用VGG16迁移学习策略可有效提高模型性能。在迁移学习的基础上增加注意力机制并改进卷积块,该模型的均交并比值进一步提升4.35%,表明改进U-Net语义分割模型在缺陷的分类与分割任务中性能更优,具有一定的有效性和实用性。
  • 中海油重大科技项目(GD2021ZCAF0021)
参考文献 引证文献
排序方式:
[1]
李志川, 胡鹏, 马佳星, 等. 中国海上风电发展现状分析及展望[J]. 中国海上油气, 2022, 34(5):229-236.
LI Zhichuan, HU Peng, MA Jiaxing, et al. Analysis and prospect of offshore wind power development in China[J]. China Offshore Oil and Gas, 2022, 34(5):229-236.
[2]
TAVNER P J, XIANG J, SPINATO F. Reliability analysis for wind turbines[J]. Wind Energy, 2007, 10(1):1-18.
[3]
YANG Wenxian, PENG Zhike, WEI Kexiang, et al. Structural health monitoring of composite wind turbine blades: challenges, issues and potential solutions[J]. IET Renewable Power Generation, 2017, 11(4): 411-416.
[4]
DU Ying, ZHOU Shengxi, JING Xingjian, et al. Damage detection techniques for wind turbine blades: a review[J]. Mechanical Systems and Signal Processing, 2020,141:DOI:10.1016/j.ymssp.2019.106445.
[5]
HU Wenbo, WANG Weidong, AI Chengbo, et al. Machine vision-based surface crack analysis for transportation infrastructure[J]. Automation in Construction, 2021,132:DOI: 10.1016/j.autcon.2021.103973.
[6]
陈法法, 成孟腾, 杨蕴鹏, 等. 融合双注意力机制和U-Net网络的锈蚀图像分割[J]. 西安交通大学学报, 2021, 55(12):119-128.
CHEN Fafa, CHENG Mengteng, YANG Yunpeng, et al. A segmentation method based on dual attention mechanism and U-Net for corrosion images[J]. Journal of Xi'an Jiaotong University, 2021, 55(12):119-128.
[7]
XU Donghua, WEN Chuanbo, LIU Jihui. Wind turbine blade surface inspection based on deep learning and UAV-taken images[J]. Renewable Sustainable Energy, 2019, 11(5):DOI:10.1063/1.5113532.
[8]
SARKAR D, GUNTURI S K. Wind turbine blade structural state evaluation by hybrid object detector relying on deep learning models[J]. Journal of Ambient Intelligence and Humanized Computing, 2021, 12(8): 8535-8548.
[9]
MAO Yulin, WANG Shuanxin, YU Dingli, et al. Automatic image detection of multi-type sur face defects on wind turbine blades based on cascade deep learning network[J]. Intelligent Data Analysis, 2021, 25(2):463-482.
[10]
SHELHAMER E, LONG J, DARRELL T. Fully convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 39(4): 640-651.
[11]
RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedical image segmentation[C]. International Conference on Medical Image Computing and Computer Assisted Intervention, 2015:234-241.
[12]
BADRINARAYANAN V, KENDALL A, CIPOLLA R. SegNet: a deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12):2481-2495.
[13]
CHEN Liangchieh, PAPANDREOU G, KOKKINOS I, et al. DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 40(4):834-848.
[14]
CUI Xiaoning, WANG Qicai, DAI Jinpeng, et al. Intelligent crack detection based on attention mechanism in convolution neural network[J]. Advances in Structural Engineering, 2021, 24(9): 1859-1868.
[15]
HACEFENDIOLU K, BASAGA H B, YAVUZ Z, et al. Intelligent ice detection on wind turbine blades using semantic segmentation and class activation map approaches based on deep learning method[J]. Renewable Energy, 2022, 182:1-16.
[16]
夏正洪, 魏汝祥, 李彦冬. 改进YOLOv3算法下通航机场场面运动目标检测[J]. 中国安全科学学报, 2023, 33(2):82-88.
XIA Zhenghong, WEI Ruxiang, LI Yandong. Moving target detection of general aviation airport based on improved YOLOv3 algorithm[J]. China Safety Science Journal, 2023, 33(2):82-88.
[17]
何静, 侯娜, 张昌凡, 等. 基于金字塔拆分注意力的列车轮对踏面损伤诊断[J]. 中国安全科学学报, 2022, 32(5):35-40.
HE Jing, HOU Na, ZHANG Changfan, et al. Diagnosis of train wheelset tread damage based on EPSA-ResNet[J]. China Safety Science Journal, 2022, 32(5):35-40.
[18]
WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]. Proceedings of the European Conference on Computer Vision (ECCV), 2018: 3-19.
[19]
张欢, 仇大伟, 冯毅博, 等. U-Net模型改进及其在医学图像分割上的研究综述[J]. 激光与光电子学进展, 2022, 59(2):1-17.
ZHANG Huan, QIU Dawei, FENG Yibo, et al. Improved U-Net models and its applications in medical image segmentation: a review[J]. Laser & Optoelectronics Progress, 2022, 59(2):1-17.
2024年第34卷第5期
PDF下载
373
151
引用本文
BibTeX
文章信息
doi: 10.16265/j.cnki.issn1003-3033.2024.05.0767
  • 接收时间:2023-11-14
  • 首发时间:2025-07-14
  • 出版时间:2024-05-28
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2023-11-14
  • 修回日期:2024-02-18
基金
中海油重大科技项目(GD2021ZCAF0021)
作者信息
    1 中海油能源发展股份有限公司 清洁能源分公司 天津 300459
    2 中国石油大学(北京) 安全与海洋工程学院,北京 102249
    3 应急管理部油气生产安全与应急技术重点实验室,北京 102249

通讯作者:

**梁 伟(1978—),男,陕西绥德人,博士,教授,博士生导师,主要从事油气管道安全检测、站场设备诊断与可靠性评估等方面的研究。E-mail:
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/zgaqkxxb/CN/10.16265/j.cnki.issn1003-3033.2024.05.0767
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