首页 期刊 论文 学科刊群 资讯 加盟 关于
EN
image
Article(id=1210613541879541954, tenantId=1146029695717560320, journalId=1149653034449285133, issueId=1210613541111984320, articleNumber=null, orderNo=null, doi=10.16790/j.cnki.1009-9239.im.2024.11.017, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=research-article, receivedDate=1712160000000, receivedDateStr=2024-04-04, revisedDate=1718899200000, revisedDateStr=2024-06-21, acceptedDate=null, acceptedDateStr=null, onlineDate=1766562360496, onlineDateStr=2025-12-24, pubDate=1732032000000, pubDateStr=2024-11-20, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1766562360496, onlineIssueDateStr=2025-12-24, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1766562360496, creator=13701087609, updateTime=1766562360496, updator=13701087609, issue=Issue{id=1210613541111984320, tenantId=1146029695717560320, journalId=1149653034449285133, year='2024', volume='57', issue='11', pageStart='1', pageEnd='143', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1766562360313, creator=13701087609, updateTime=1766564044072, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1210620603392987671, tenantId=1146029695717560320, journalId=1149653034449285133, issueId=1210613541111984320, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1210620603392987672, tenantId=1146029695717560320, journalId=1149653034449285133, issueId=1210613541111984320, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=135, endPage=143, ext={EN=ArticleExt(id=1210613542135394502, articleId=1210613541879541954, tenantId=1146029695717560320, journalId=1149653034449285133, language=EN, title=Research on semantic segmentation and recognition algorithm integrating multi-scale convolutional attention mechanism for algae pollution on insulator surface, columnId=1192878364340924664, journalTitle=Insulating Materials, columnName=Test and Analysis, runingTitle=null, highlight=null, articleAbstract=

Algae is a special kind of biofouling, its attachment on the insulator surface of electrical equipment will significantly reduce the fouling flash resistance of insulator , which poses a threat to the safe and stable operation of power grids. In this paper, a semantic segmentation algorithm of insulating algae integrated multi-scale convolution attention mechanism was proposed. Firstly, a model for insulating algae class semantic segmentation was constructed on the basic U-Net network model, and VGG16 was used as the backbone feature extraction network. The model adopts a U-shaped structure, and the left side is the feature extraction part of the VGG16 backbone, which can effectively extract the informations of five feature layers. The right side was the enhanced feature extraction part. CBAM module was selected for attention module, and the multi-scale convolution was introduced based on CBAM module. Then the CABM convolutional attention module was added to the encoder and the decoder of U-Net network before up-sampling and down-sampling. Finally, the model was compared with Deeplabv3+ and Transfuse network on the self-constructed algae-covered insulator image dataset. The results show that compared with the basic U-Net model, the mIoU value of this model improves by 0.28, mPA value improves by 0.27, Dice coefficient improves by 0.06, Hausdorff distance reduces by 11.77, and the RVE value reduces by 0.06. The visualization results of the segmentation process demonstrates that the model in this paper can pay more attention to the algal coverage region, and locate the boundary of the algal coverage region more accurately, which reduces the segmentation error effectively.

, 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=Shifang YANG, Xinyi ZANG, Mingxi ZHU, Yunpeng LIU, Chaojun SHI, Zhidong JIA), CN=ArticleExt(id=1210613546073846132, articleId=1210613541879541954, tenantId=1146029695717560320, journalId=1149653034449285133, language=CN, title=融合多尺度卷积注意力机制的绝缘子表面藻类污秽语义分割和识别算法研究, columnId=1192878364483531003, journalTitle=绝缘材料, columnName=测试与分析, runingTitle=null, highlight=null, articleAbstract=

藻类是一种特殊生物污秽，其在电气设备绝缘子表面附着会使绝缘子耐污闪能力大幅下降，给电网安全稳定运行带来威胁。本文提出了一种融合多尺度卷积注意力机制的绝缘藻类语义分割算法。首先在基础U-Net网络模型上构建了一种用于绝缘藻类等级语义分割的模型，并采用VGG16作为主干特征提取网络。模型采用U型结构，左侧为VGG16及主干特征提取部分，能够有效提取5个特征层信息；右侧为加强特征提取部分。注意力模块选取CBAM模块，并在其基础上引入多尺度卷积；随后在U-Net网络的编码器和解码器进行上采样与下采样之前均加入CBAM卷积注意力模块。最后将本文模型在自行构建的藻类覆盖绝缘子图像数据集上与Deeplabv3+与Transfuse网络进行对比。结果表明：相比于基础U-Net模型，本文模型mIoU值提高了0.28，mPA值提高了0.27，Dice系数提高了0.06，豪斯多夫距离降低了11.77，RVE值降低了0.06。分割过程可视化结果展示了本文模型能更好地关注藻类覆盖区域，且对藻类覆盖区域边界的定位更准确，有效减小了分割误差。

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=M8PEGYn8hmZ3fKVlHiiiHA==, magXml=2qmrV30ZWB8qhgucZzLCzw==, pdfUrl=null, pdf=hYoWUIBDAgbik/BJS2rg1w==, pdfFileSize=9293152, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=kkMh/qrKYve5YtAAAxYB8A==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=//7hAlJJkZx/SkvdfBOHsQ==, mapNumber=null, authorCompany=null, fund=null, authors=

杨世芳（1994-），女（汉族），河北保定人，讲师，主要从事高电压绝缘生物染污问题、退役绝缘材料无害化处理等方面的研究。

, authorsList=杨世芳, 臧欣怡, 朱明曦, 刘云鹏, 石超君, 贾志东)}, authors=[Author(id=1218964276220449432, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, orderNo=0, 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=1218964276337889948, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964276220449432, language=EN, stringName=Shifang YANG, firstName=Shifang, middleName=null, lastName=YANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1Department of Electric Power Engineering, North China Electric Power University, Baoding 071000, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1218964277675872934, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964276220449432, language=CN, stringName=杨世芳, firstName=世芳, middleName=null, lastName=杨, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1华北电力大学，电力工程系，河北 保定 071000, bio={"content":"

杨世芳（1994-），女（汉族），河北保定人，讲师，主要从事高电压绝缘生物染污问题、退役绝缘材料无害化处理等方面的研究。

"}, bioImg=null, bioContent=

杨世芳（1994-），女（汉族），河北保定人，讲师，主要从事高电压绝缘生物染污问题、退役绝缘材料无害化处理等方面的研究。

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1218964275641635430, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, xref=1, ext=[AuthorCompanyExt(id=1218964275650024039, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275641635430, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Department of Electric Power Engineering, North China Electric Power University, Baoding 071000, China), AuthorCompanyExt(id=1218964275654218344, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275641635430, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1华北电力大学，电力工程系，河北 保定 071000)])]), Author(id=1218964277835256491, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, 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=1218964277982057138, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964277835256491, language=EN, stringName=Xinyi ZANG, firstName=Xinyi, middleName=null, lastName=ZANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1Department of Electric Power Engineering, North China Electric Power University, Baoding 071000, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1218964278099497656, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964277835256491, language=CN, stringName=臧欣怡, firstName=欣怡, middleName=null, lastName=臧, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1华北电力大学，电力工程系，河北 保定 071000, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1218964275641635430, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, xref=1, ext=[AuthorCompanyExt(id=1218964275650024039, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275641635430, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Department of Electric Power Engineering, North China Electric Power University, Baoding 071000, China), AuthorCompanyExt(id=1218964275654218344, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275641635430, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1华北电力大学，电力工程系，河北 保定 071000)])]), Author(id=1218964278221132478, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, orderNo=2, 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=1218964278439236296, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964278221132478, language=EN, stringName=Mingxi ZHU, firstName=Mingxi, middleName=null, lastName=ZHU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=3, address=3State Grid Shaanxi Electric Power Research Institute, Xi′an 710100, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1218964278560871117, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964278221132478, language=CN, stringName=朱明曦, firstName=明曦, middleName=null, lastName=朱, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=3, address=3国网陕西省电力公司电力科学研究院，陕西 西安 710100, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1218964275960402559, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, xref=3, ext=[AuthorCompanyExt(id=1218964275972985471, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275960402559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3State Grid Shaanxi Electric Power Research Institute, Xi′an 710100, China), AuthorCompanyExt(id=1218964275985568385, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275960402559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3国网陕西省电力公司电力科学研究院，陕西 西安 710100)])]), Author(id=1218964278657340114, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, 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=1218964278762197720, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964278657340114, language=EN, stringName=Yunpeng LIU, firstName=Yunpeng, middleName=null, lastName=LIU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1Department of Electric Power Engineering, North China Electric Power University, Baoding 071000, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1218964278854472418, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964278657340114, language=CN, stringName=刘云鹏, firstName=云鹏, middleName=null, lastName=刘, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1华北电力大学，电力工程系，河北 保定 071000, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1218964275641635430, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, xref=1, ext=[AuthorCompanyExt(id=1218964275650024039, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275641635430, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Department of Electric Power Engineering, North China Electric Power University, Baoding 071000, China), AuthorCompanyExt(id=1218964275654218344, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275641635430, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1华北电力大学，电力工程系，河北 保定 071000)])]), Author(id=1218964278955135721, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, 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=1218964279068381936, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964278955135721, language=EN, stringName=Chaojun SHI, firstName=Chaojun, middleName=null, lastName=SHI, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2Department of Electronic and Communication Engineering, North China Electric Power University, Baoding 071000, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1218964279244542713, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964278955135721, language=CN, stringName=石超君, firstName=超君, middleName=null, lastName=石, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2华北电力大学，电子与通信工程系，河北 保定 071000, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1218964275771658866, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, xref=2, ext=[AuthorCompanyExt(id=1218964275775853171, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275771658866, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2Department of Electronic and Communication Engineering, North China Electric Power University, Baoding 071000, China), AuthorCompanyExt(id=1218964275784241780, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275771658866, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2华北电力大学，电子与通信工程系，河北 保定 071000)])]), Author(id=1218964279387149060, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, orderNo=5, 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=1218964279496200974, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964279387149060, language=EN, stringName=Zhidong JIA, firstName=Zhidong, middleName=null, lastName=JIA, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=4, address=4Tsinghua Shenzhen International Graduate School, Shenzhen 518055, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1218964279596864277, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, authorId=1218964279387149060, language=CN, stringName=贾志东, firstName=志东, middleName=null, lastName=贾, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=4, address=4清华大学深圳国际研究生院，广东 深圳 518055, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1218964276086231691, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, xref=4, ext=[AuthorCompanyExt(id=1218964276094620302, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964276086231691, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4Tsinghua Shenzhen International Graduate School, Shenzhen 518055, China), AuthorCompanyExt(id=1218964276103008910, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964276086231691, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4清华大学深圳国际研究生院，广东 深圳 518055)])])], keywords=[Keyword(id=1218964279798190880, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, orderNo=1, keyword=insulating algae), Keyword(id=1218964280028877612, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, orderNo=2, keyword=image processing), Keyword(id=1218964280142123823, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, orderNo=3, keyword=semantic segmentation), Keyword(id=1218964280251175735, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, orderNo=4, keyword=attention mechanisms), Keyword(id=1218964280356033343, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, orderNo=1, keyword=绝缘藻类), Keyword(id=1218964280435725123, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, orderNo=2, keyword=图象处理), Keyword(id=1218964280532194120, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, orderNo=3, keyword=语义分割), Keyword(id=1218964280624468815, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, orderNo=4, keyword=注意力机制)], refs=[Reference(id=1218964285187871726, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2010, volume=34, issue=12, pageStart=124, pageEnd=130, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=宿志一, 李庆峰, journalName=电网技术, refType=null, unstructuredReference=宿志一,李庆峰.我国电网防污闪措施的回顾和总结[J].电网技术,2010,34(12):124-130., articleTitle=我国电网防污闪措施的回顾和总结, refAbstract=null), Reference(id=1218964285275952114, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2010, volume=34, issue=12, pageStart=124, pageEnd=130, url=null, language=null, rfNumber=[1], rfOrder=1, authorNames=SUZ Y, LIQ F, journalName=Power System Technology, refType=null, unstructuredReference=SUZ Y, LIQ F. Historical review and summary on measures against pollution flashover occurred in power grids in China[J]. Power System Technology,2010,34(12):124-130., articleTitle=Historical review and summary on measures against pollution flashover occurred in power grids in China, refAbstract=null), Reference(id=1218964285364032506, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2023, volume=35, issue=16, pageStart=11833, pageEnd=11846, url=null, language=null, rfNumber=[2], rfOrder=2, authorNames=AGARWALM, GUPTAS K, BISWASK K, journalName=Neural computing & applications, refType=null, unstructuredReference=AGARWALM, GUPTAS K, BISWASK K. Development of a compressed FCN architecture for semantic segmentation using particle swarm optimization[J]. Neural computing & applications,2023,35(16):11833-11846., articleTitle=Development of a compressed FCN architecture for semantic segmentation using particle swarm optimization, refAbstract=null), Reference(id=1218964286689432574, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2000, volume=15, issue=4, pageStart=335, pageEnd=360, url=null, language=null, rfNumber=[3], rfOrder=3, authorNames=FERNANDOM, GUBANSKIS, journalName=IEEE Transactions on Power Delivery, refType=null, unstructuredReference=FERNANDOM, GUBANSKIS. Closure to discussion of performance of nonceramic insulators under tropical field conditions [J]. IEEE Transactions on Power Delivery,2000,15(4):335-360., articleTitle=Closure to discussion of performance of nonceramic insulators under tropical field conditions, refAbstract=null), Reference(id=1218964286777511936, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=1998, volume=59, issue=1, pageStart=309, pageEnd=315, url=null, language=null, rfNumber=[4], rfOrder=4, authorNames=FLEMMINGlH C, journalName=Polymer Degradation & Stability, refType=null, unstructuredReference=FLEMMINGlH C. Relevance of biofilms for the biodeterioration of surfaces of polymeric materials[J]. Polymer Degradation & Stability,1998,59(1):309-315., articleTitle=Relevance of biofilms for the biodeterioration of surfaces of polymeric materials, refAbstract=null), Reference(id=1218964286882369542, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2017, volume=28, issue=2, pageStart=3, pageEnd=14, url=null, language=null, rfNumber=[5], rfOrder=5, authorNames=ROJASH E, ROJASH D, journalName=Información tecnológica, refType=null, unstructuredReference=ROJASH E, ROJASH D. Performance evaluation of ceramic and polymeric electrical distribution insulators under fungus biocontamination[J]. Información tecnológica,2017,28(2):3-14., articleTitle=Performance evaluation of ceramic and polymeric electrical distribution insulators under fungus biocontamination, refAbstract=null), Reference(id=1218964286978838539, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=1998, volume=13, issue=2, pageStart=316, pageEnd=322, url=null, language=null, rfNumber=[6], rfOrder=6, authorNames=GORURR S, BERNSTEINB S, journalName=IEEE Transactions on Power Delivery, refType=null, unstructuredReference=GORURR S, BERNSTEINB S. Field and laboratory aging of polymeric distribution cable terminations. I. field agin [J]. IEEE Transactions on Power Delivery,1998,13(2):316-322., articleTitle=Field and laboratory aging of polymeric distribution cable terminations. I. field agin, refAbstract=null), Reference(id=1218964287096279056, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=1988, volume=15, issue=3, pageStart=215, pageEnd=218, url=null, language=null, rfNumber=[7], rfOrder=7, authorNames=COOWARF, HOOLEP R P, journalName=Electric Power Systems Research, refType=null, unstructuredReference=COOWARF, HOOLEP R P. Discharge characteristics of alga-contaminated string insulators in the tropics[J]. Electric Power Systems Research,1988,15(3):215-218., articleTitle=Discharge characteristics of alga-contaminated string insulators in the tropics, refAbstract=null), Reference(id=1218964287192748049, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2021, volume=7, issue=9, pageStart=210, pageEnd=217, url=null, language=null, rfNumber=[8], rfOrder=8, authorNames=GUH, MINZ, WENW, journalName=Energy Reports, refType=null, unstructuredReference=GUH, MINZ, WENW, et al. Improved U-Net based insulator image segmentation method based on attention mechanism[J]. Energy Reports,2021,7(9):210-217., articleTitle=Improved U-Net based insulator image segmentation method based on attention mechanism, refAbstract=null), Reference(id=1218964287310188569, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=1990, volume=5, issue=3, pageStart=1634, pageEnd=1640, url=null, language=null, rfNumber=[9], rfOrder=9, authorNames=NAITOK, MATSUNKAR, SAKANISHIK, journalName=IEEE Transactions on Power Delivery, refType=null, unstructuredReference=NAITOK, MATSUNKAR, SAKANISHIK. Investigation of the insulation performance of the insulator covered with lichen[J]. IEEE Transactions on Power Delivery,1990,5(3):1634-1640., articleTitle=Investigation of the insulation performance of the insulator covered with lichen, refAbstract=null), Reference(id=1218964287402463262, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2020, volume=8, issue=5, pageStart=133882, pageEnd=133889, url=null, language=null, rfNumber=[10], rfOrder=10, authorNames=HEH, HUZ, WANGB, journalName=IEEE Access, refType=null, unstructuredReference=HEH, HUZ, WANGB, et al. A contactless zero-value insulators detection method based on infrared images matching[J]. IEEE Access,2020,8(5):133882-133889., articleTitle=A contactless zero-value insulators detection method based on infrared images matching, refAbstract=null), Reference(id=1218964287528292387, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=1996, volume=42, issue=1, pageStart=35, pageEnd=39, url=null, language=null, rfNumber=[11], rfOrder=11, authorNames=ROBERTD M, ROBERTD H, JOEM K, journalName=Electric Power Systems Research, refType=null, unstructuredReference=ROBERTD M, ROBERTD H, JOEM K, et al. A study of biological contaminants on high voltage porcelain insulators[J]. Electric Power Systems Research,1996,42(1):35-39., articleTitle=A study of biological contaminants on high voltage porcelain insulators, refAbstract=null), Reference(id=1218964287641538599, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2019, volume=7, issue=8, pageStart=165710, pageEnd=165723, url=null, language=null, rfNumber=[12], rfOrder=12, authorNames=YIL, LiZ, LIJ, journalName=IEEE Access, refType=null, unstructuredReference=YIL, LiZ, LIJ, et al. Fault diagnosis of rotating machinery based on combination of deep belief network and one-dimensional convolutional neural network[J]. IEEE Access,2019,7(8):165710-165723., articleTitle=Fault diagnosis of rotating machinery based on combination of deep belief network and one-dimensional convolutional neural network, refAbstract=null), Reference(id=1218964287721230378, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2021, volume=14, issue=20, pageStart=6662, pageEnd=6668, url=null, language=null, rfNumber=[13], rfOrder=13, authorNames=DA Z, SHUAIL C, journalName=Energies, refType=null, unstructuredReference=DA Z, SHUAIL C. Insulator contamination grade recognition using the deep learning of color information of images[J]. Energies,2021,14(20):6662-6668., articleTitle=Insulator contamination grade recognition using the deep learning of color information of images, refAbstract=null), Reference(id=1218964287800922160, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2010, volume=30, issue=4, pageStart=132, pageEnd=138, url=null, language=null, rfNumber=[14], rfOrder=14, authorNames=李佐胜, 李文利, 姚建刚, journalName=中国电机工程学报, refType=null, unstructuredReference=李佐胜,李文利,姚建刚,等.应用绝缘子红外热像处理的现场污秽等级检测方法 [J].中国电机工程学报,2010,30(4):132-138., articleTitle=应用绝缘子红外热像处理的现场污秽等级检测方法, refAbstract=null), Reference(id=1218964287889002548, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2010, volume=30, issue=4, pageStart=132, pageEnd=138, url=null, language=null, rfNumber=[14], rfOrder=15, authorNames=LIZ S, LIW L, YAOJ G, journalName=Proceedings of the CSEE, refType=null, unstructuredReference=LIZ S, LIW L, YAOJ G, et al. On-site detection of pollution level of insulators based on infrared-thermal-image processing[J]. Proceedings of the CSEE,2010,30(4):132-138., articleTitle=On-site detection of pollution level of insulators based on infrared-thermal-image processing, refAbstract=null), Reference(id=1218964287993860156, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2021, volume=37, issue=8, pageStart=116, pageEnd=128, url=null, language=null, rfNumber=[15], rfOrder=16, authorNames=WANGZ, AFGANM S, GUW, journalName=TrAC Trends in Analytical Chemistry, refType=null, unstructuredReference=WANGZ, AFGANM S, GUW, et al. Recent advances in laser-induced breakdown spectroscopy quantification: from fundamental understanding to data processing[J]. TrAC Trends in Analytical Chemistry,2021,37(8):116-128., articleTitle=Recent advances in laser-induced breakdown spectroscopy quantification: from fundamental understanding to data processing, refAbstract=null), Reference(id=1218964288077746236, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2023, volume=34, issue=5, pageStart=44, pageEnd=48, url=null, language=null, rfNumber=[16], rfOrder=17, authorNames=GUOH L, QIANW M, journalName=Measurement Science and Technology, refType=null, unstructuredReference=GUOH L, QIANW M. Strip steel surface defect detecting method combined with a multi-layer attention mechanism network[J]. Measurement Science and Technology,2023,34(5):44-48., articleTitle=Strip steel surface defect detecting method combined with a multi-layer attention mechanism network, refAbstract=null), Reference(id=1218964288195186752, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2015, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[17], rfOrder=18, authorNames=RONNEBERGERO, FISCHERP, BROXT, journalName=null, refType=null, unstructuredReference=RONNEBERGERO, FISCHERP, BROXT. U-Net: convolutional net-works for biomedical image segmentation[C]//Proceedings of the 18th International Conference on Medical Image Computing and Computer-Assisted Intervention. Munich, Germany:Springer,2015., articleTitle=U-Net: convolutional net-works for biomedical image segmentation, refAbstract=null), Reference(id=1218964288341987396, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2020, volume=57, issue=4, pageStart=170, pageEnd=180, url=null, language=null, rfNumber=[18], rfOrder=19, authorNames=刘航, 汪西莉, journalName=激光与光电子学进展, refType=null, unstructuredReference=刘航,汪西莉.基于注意力机制的遥感图像分割模型[J].激光与光电子学进展,2020,57(4):170-180., articleTitle=基于注意力机制的遥感图像分割模型, refAbstract=null), Reference(id=1218964288421679175, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2020, volume=57, issue=4, pageStart=170, pageEnd=180, url=null, language=null, rfNumber=[18], rfOrder=20, authorNames=LIUH, WANGX L, journalName=Laser and Optoelectronics Progress, refType=null, unstructuredReference=LIUH, WANGX L. Remote sensing image segmentation model based on attention mechanism[J]. Laser and Optoelectronics Progress,2020,57(4):170-180, articleTitle=Remote sensing image segmentation model based on attention mechanism, refAbstract=null), Reference(id=1218964288534925388, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2017, volume=39, issue=4, pageStart=640, pageEnd=651, url=null, language=null, rfNumber=[19], rfOrder=21, authorNames=EVANS, JONATHANL, TREVORD, journalName=IEEE transactions on pattern analysis and machine intelligence, refType=null, unstructuredReference=EVANS, JONATHANL, TREVORD. Fully Convolutional Networks for Semantic Segmentation[J]. IEEE transactions on pattern analysis and machine intelligence,2017,39(4):640-651., articleTitle=Fully Convolutional Networks for Semantic Segmentation, refAbstract=null), Reference(id=1218964288610422865, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2023, volume=12, issue=3, pageStart=487, pageEnd=497, url=null, language=null, rfNumber=[20], rfOrder=22, authorNames=ZIFEIL, WENZ Y, RURUG, journalName=Electronics, refType=null, unstructuredReference=ZIFEIL, WENZ Y, RURUG, et al. Transattention U-Net for semantic segmentation of poppy[J]. Electronics,2023,12(3):487-497., articleTitle=Transattention U-Net for semantic segmentation of poppy, refAbstract=null), Reference(id=1218964288711086164, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2020, volume=106, issue=5, pageStart=665, pageEnd=675, url=null, language=null, rfNumber=[21], rfOrder=23, authorNames=QINX, ZHANGZ, HUANC, journalName=Pattern Recognition, refType=null, unstructuredReference=QINX, ZHANGZ, HUANC, et al. U2-Net: going deeper with nested U-structure for salient object detection[J]. Pattern Recognition,2020,106(5):665-675., articleTitle=U2-Net: going deeper with nested U-structure for salient object detection, refAbstract=null), Reference(id=1218964288782389336, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2023, volume=208, issue=6, pageStart=55, pageEnd=62, url=null, language=null, rfNumber=[22], rfOrder=24, authorNames=JINGS, JIE X, ALLEN, journalName=Measurement, refType=null, unstructuredReference=JINGS, JIE X, ALLEN. Z A,et al. Automatic pixel-level pavement sealed crack detection using multi-fusion U-Net network[J]. Measurement,2023,208(6):55-62., articleTitle=Z A,, refAbstract=null), Reference(id=1218964288866275419, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2023, volume=171, issue=5, pageStart=104984, pageEnd=104994, url=null, language=null, rfNumber=[23], rfOrder=25, authorNames=QIZ, YINGL, YIZ, journalName=International Journal of Medical Informatics, refType=null, unstructuredReference=QIZ, YINGL, YIZ, et al. A comparative study of attention mechanism based deep learning methods for bladder tumor segmentation[J]. International Journal of Medical Informatics,2023, 171(5):104984-104994., articleTitle=A comparative study of attention mechanism based deep learning methods for bladder tumor segmentation, refAbstract=null), Reference(id=1218964288929189983, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2024, volume=47, issue=1, pageStart=148, pageEnd=171, url=null, language=null, rfNumber=[24], rfOrder=26, authorNames=肖国庆, 李雪琪, 陈玥丹, journalName=计算机学报, refType=null, unstructuredReference=肖国庆,李雪琪,陈玥丹,等.大规模图神经网络研究综述[J].计算机学报,2024,47(1):148-171., articleTitle=大规模图神经网络研究综述, refAbstract=null), Reference(id=1218964289008881762, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2024, volume=47, issue=1, pageStart=148, pageEnd=171, url=null, language=null, rfNumber=[24], rfOrder=27, authorNames=XIAOG Q, LIX Q, CHENY D, journalName=Chinese Journal of Computers, refType=null, unstructuredReference=XIAOG Q, LIX Q, CHENY D. A survey of large-scale graph neural networks[J]. Chinese Journal of Computers,2024,47(1):148-171., articleTitle=A survey of large-scale graph neural networks, refAbstract=null), Reference(id=1218964289084379236, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2021, volume=7, issue=z7, pageStart=187, pageEnd=197, url=null, language=null, rfNumber=[25], rfOrder=28, authorNames=HANG J, HEM, ZHAOF, journalName=Energy Reports, refType=null, unstructuredReference=HANG J, HEM, ZHAOF, et al. Insulator detection and damage identification based on improved lightweight YOLOv4 network[J]. Energy Reports,2021,7(z7):187-197., articleTitle=Insulator detection and damage identification based on improved lightweight YOLOv4 network, refAbstract=null), Reference(id=1218964289155682409, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2020, volume=59, issue=7, pageStart=1907, pageEnd=1915, url=null, language=null, rfNumber=[26], rfOrder=29, authorNames=WANGJ, ZHOUY, YANGY, journalName=Applied Optics, refType=null, unstructuredReference=WANGJ, ZHOUY, YANGY. Rapid 3D measurement technique for colorful objects employing RGB color light projection[J]. Applied Optics,2020,59(7):1907-1915., articleTitle=Rapid 3D measurement technique for colorful objects employing RGB color light projection, refAbstract=null), Reference(id=1218964289231179886, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[27], rfOrder=30, authorNames=WOO S, PARKJ, LEEJ Y, journalName=null, refType=null, unstructuredReference=WOO S, PARKJ, LEEJ Y, et al. CBAM: convolutional block attention module[C]//Proceedings of the 15th European Conference on Computer Vision (ECCV). Munich, Germany:Springer,2018., articleTitle=CBAM: convolutional block attention module, refAbstract=null), Reference(id=1218964289331843187, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2020, volume=4, issue=9, pageStart=1, pageEnd=9, url=null, language=null, rfNumber=[28], rfOrder=31, authorNames=WANGJ H, ZHOUY G, YANGY X, journalName=IEEE Sensors Journal, refType=null, unstructuredReference=WANGJ H, ZHOUY G, YANGY X. Three-dimensional shape detection for non uniform reflective objects: combination of color light projection and camera's exposure adjustment[J]. IEEE Sensors Journal,2020,4(9):1-9., articleTitle=Three-dimensional shape detection for non uniform reflective objects: combination of color light projection and camera's exposure adjustment, refAbstract=null), Reference(id=1218964289419923575, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2018, volume=33, issue=4, pageStart=222, pageEnd=231, url=null, language=null, rfNumber=[29], rfOrder=32, authorNames=OKTAYO, SCHLEMPERJ, FOLGOCL L, journalName=Computer Vision and Pattern Recognition, refType=null, unstructuredReference=OKTAYO, SCHLEMPERJ, FOLGOCL L, et al. Attention U-net: learning where to look for the pancreas[J]. Computer Vision and Pattern Recognition,2018,33(4):222-231., articleTitle=Attention U-net: learning where to look for the pancreas, refAbstract=null), Reference(id=1218964289508003959, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2023, volume=13, issue=3, pageStart=1685, pageEnd=1695, url=null, language=null, rfNumber=[30], rfOrder=33, authorNames=KAI W, ZHONGZ, CHAOZ C, journalName=Applied Sciences, refType=null, unstructuredReference=KAI W, ZHONGZ, CHAOZ C, et al. Intelligent whistling system of rail train based on YOLOv4 and U-net[J]. Applied Sciences,2023,13(3):1685-1695., articleTitle=Intelligent whistling system of rail train based on YOLOv4 and U-net, refAbstract=null), Reference(id=1218964289579307130, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2020, volume=11, issue=2, pageStart=125, pageEnd=131, url=null, language=null, rfNumber=[31], rfOrder=34, authorNames=BUSLARVA, IGLOVIKOVV I, KHVEDCHENYAE, journalName=Information, refType=null, unstructuredReference=BUSLARVA, IGLOVIKOVV I, KHVEDCHENYAE, et al. Albumentations: fast and flexible image augmentations[J]. Information,2020,11(2):125-131., articleTitle=Albumentations: fast and flexible image augmentations, refAbstract=null), Reference(id=1218964289650610300, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2019, volume=17, issue=2, pageStart=22, pageEnd=28, url=null, language=null, rfNumber=[32], rfOrder=35, authorNames=YANB, CHENQ, YER, journalName=International Journal of Wavelets, Multiresolution and Information Processing, refType=null, unstructuredReference=YANB, CHENQ, YER, et al. Insulator detection and recognition of explosion based on convolutional neural networks[J]. International Journal of Wavelets, Multiresolution and Information Processing,2019,17(2):22-28., articleTitle=Insulator detection and recognition of explosion based on convolutional neural networks, refAbstract=null), Reference(id=1218964289730302079, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2022, volume=22, issue=1, pageStart=44, pageEnd=51, url=null, language=null, rfNumber=[33], rfOrder=36, authorNames=YAOX, LEIL W, ZHENGZ, journalName=Journal of Physics: Conference Series, refType=null, unstructuredReference=YAOX, LEIL W, ZHENGZ, et al. Consistency-based semi-supervised learning framework for power line insulators detection[J]. Journal of Physics: Conference Series,2022,22(1):44-51., articleTitle=Consistency-based semi-supervised learning framework for power line insulators detection, refAbstract=null), Reference(id=1218964289835159682, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2022, volume=55, issue=9, pageStart=100, pageEnd=106, url=null, language=null, rfNumber=[34], rfOrder=37, authorNames=杨秋玉, 郑小刚, 李建兴, journalName=绝缘材料, refType=null, unstructuredReference=杨秋玉,郑小刚,李建兴,等.基于图像处理的复合绝缘子憎水性智能识别方法[J].绝缘材料,2022,55(9):100-106., articleTitle=基于图像处理的复合绝缘子憎水性智能识别方法, refAbstract=null), Reference(id=1218964291110228101, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2022, volume=55, issue=9, pageStart=100, pageEnd=106, url=null, language=null, rfNumber=[34], rfOrder=38, authorNames=YANGQ Y, ZHENGX G, LIJ X, journalName=Insulating Materials, refType=null, unstructuredReference=YANGQ Y, ZHENGX G, LIJ X, et al. Intelligent recognition method of composite insulator hydrophnbicity based on image processing[J]. Insulating Materials,2022,55(9):100-106., articleTitle=Intelligent recognition method of composite insulator hydrophnbicity based on image processing, refAbstract=null), Reference(id=1218964291198308487, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2021, volume=54, issue=7, pageStart=10, pageEnd=24, url=null, language=null, rfNumber=[35], rfOrder=39, authorNames=范路, 陆云才, 陶风波, journalName=绝缘材料, refType=null, unstructuredReference=范路,陆云才,陶风波,等.人工智能在局部放电检测中的应用(二):模式识别与状态评估[J].绝缘材料,2021,54(7):10-24., articleTitle=人工智能在局部放电检测中的应用(二):模式识别与状态评估, refAbstract=null), Reference(id=1218964291311554698, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, doi=null, pmid=null, pmcid=null, year=2021, volume=54, issue=7, pageStart=10, pageEnd=24, url=null, language=null, rfNumber=[35], rfOrder=40, authorNames=FANL, LUY C, TAOF B, journalName=Insulating Materials, refType=null, unstructuredReference=FANL, LUY C, TAOF B, et. al. Application of artificial intelligence in partial discharge detection part Ⅱ: pattern recognition and condition assessment[J]. Insulating Materials,2021,54(7):10-24., articleTitle=. Application of artificial intelligence in partial discharge detection part Ⅱ: pattern recognition and condition assessment, refAbstract=null)], funds=[Fund(id=1218964284785218526, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, awardId=52207154, language=CN, fundingSource=国家自然科学基金资助项目(52207154), fundOrder=null, country=null), Fund(id=1218964284860716003, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, awardId=E2023502087, language=CN, fundingSource=河北省自然科学基金资助项目(E2023502087), fundOrder=null, country=null), Fund(id=1218964284944602087, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, awardId=2023MS104, language=CN, fundingSource=中央高校基本科研业务费专项资金资助项目(2023MS104), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1218964275641635430, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, xref=1, ext=[AuthorCompanyExt(id=1218964275650024039, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275641635430, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Department of Electric Power Engineering, North China Electric Power University, Baoding 071000, China), AuthorCompanyExt(id=1218964275654218344, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275641635430, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1华北电力大学，电力工程系，河北 保定 071000)]), AuthorCompany(id=1218964275771658866, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, xref=2, ext=[AuthorCompanyExt(id=1218964275775853171, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275771658866, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2Department of Electronic and Communication Engineering, North China Electric Power University, Baoding 071000, China), AuthorCompanyExt(id=1218964275784241780, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275771658866, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2华北电力大学，电子与通信工程系，河北 保定 071000)]), AuthorCompany(id=1218964275960402559, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, xref=3, ext=[AuthorCompanyExt(id=1218964275972985471, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275960402559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3State Grid Shaanxi Electric Power Research Institute, Xi′an 710100, China), AuthorCompanyExt(id=1218964275985568385, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964275960402559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3国网陕西省电力公司电力科学研究院，陕西 西安 710100)]), AuthorCompany(id=1218964276086231691, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, xref=4, ext=[AuthorCompanyExt(id=1218964276094620302, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964276086231691, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4Tsinghua Shenzhen International Graduate School, Shenzhen 518055, China), AuthorCompanyExt(id=1218964276103008910, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, companyId=1218964276086231691, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4清华大学深圳国际研究生院，广东 深圳 518055)])], figs=[ArticleFig(id=1218964280834184030, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Fig.1, caption=VGG16 network structure, figureFileSmall=yFmSD0uIa6WCVUiJEd0xoQ==, figureFileBig=omgxD7FrK90qlrs1a7GSyQ==, tableContent=null), ArticleFig(id=1218964282142806888, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=图1, caption=VGG16网络结构, figureFileSmall=yFmSD0uIa6WCVUiJEd0xoQ==, figureFileBig=omgxD7FrK90qlrs1a7GSyQ==, tableContent=null), ArticleFig(id=1218964282390270835, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Fig.2, caption=Feature fusion process, figureFileSmall=iBPpvejW6uJtXmd4JPNTIA==, figureFileBig=8N4arUdWFYhekieF/2ORcQ==, tableContent=null), ArticleFig(id=1218964282490934137, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=图2, caption=特征融合过程, figureFileSmall=iBPpvejW6uJtXmd4JPNTIA==, figureFileBig=8N4arUdWFYhekieF/2ORcQ==, tableContent=null), ArticleFig(id=1218964282646123390, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Fig.3, caption=CBAM module structure, figureFileSmall=Sg/BU7ar+umX90usGMJ5Lg==, figureFileBig=2pPVEJ4ACRsSfIHagz9nuA==, tableContent=null), ArticleFig(id=1218964282788729731, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=图3, caption=CBAM模块结构, figureFileSmall=Sg/BU7ar+umX90usGMJ5Lg==, figureFileBig=2pPVEJ4ACRsSfIHagz9nuA==, tableContent=null), ArticleFig(id=1218964282876810120, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Fig.4, caption=M-UNet algorithm network structure, figureFileSmall=q7YKut1IdTDZu8tcR9eGUA==, figureFileBig=u91tW77D/G1uMb2o8xhmjQ==, tableContent=null), ArticleFig(id=1218964283031999374, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=图4, caption=M-UNet算法网络结构, figureFileSmall=q7YKut1IdTDZu8tcR9eGUA==, figureFileBig=u91tW77D/G1uMb2o8xhmjQ==, tableContent=null), ArticleFig(id=1218964283128468369, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Fig.5, caption=Algae-covered insulator label diagram, figureFileSmall=623hxovVB7arLtBmGwDGDQ==, figureFileBig=1k8iv/zYeg6qZqju+y/j1A==, tableContent=null), ArticleFig(id=1218964283224937366, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=图5, caption=藻类覆盖绝缘子标签图, figureFileSmall=623hxovVB7arLtBmGwDGDQ==, figureFileBig=1k8iv/zYeg6qZqju+y/j1A==, tableContent=null), ArticleFig(id=1218964283371738011, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Fig.6, caption=Improved classification process of insulator algae coverage in U-Net model, figureFileSmall=4+KJ0TZmLIU5Hnze3mbuuA==, figureFileBig=c+KI+XhbzS5speq78lPrbA==, tableContent=null), ArticleFig(id=1218964283468207009, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=图6, caption=改进U-Net模型的绝缘子藻类覆盖等级分割流程, figureFileSmall=4+KJ0TZmLIU5Hnze3mbuuA==, figureFileBig=c+KI+XhbzS5speq78lPrbA==, tableContent=null), ArticleFig(id=1218964283598230437, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Fig.7, caption=Loss function variation diagram, figureFileSmall=L8eGhv60NU3KJ2+EVOUTSg==, figureFileBig=I4sSVEXqC28QJC0rWHzumQ==, tableContent=null), ArticleFig(id=1218964283707282346, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=图7, caption=损失函数变化图, figureFileSmall=L8eGhv60NU3KJ2+EVOUTSg==, figureFileBig=I4sSVEXqC28QJC0rWHzumQ==, tableContent=null), ArticleFig(id=1218964283770196911, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Fig.8, caption=Results of algal coverage image segmentation in different networks, figureFileSmall=HNXt0ryveV2a0K1I0MwAUQ==, figureFileBig=qRC6+rgrKX5ER86qL0D8pQ==, tableContent=null), ArticleFig(id=1218964283837305779, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=图8, caption=不同网络藻类覆盖图像分割结果, figureFileSmall=HNXt0ryveV2a0K1I0MwAUQ==, figureFileBig=qRC6+rgrKX5ER86qL0D8pQ==, tableContent=null), ArticleFig(id=1218964283908608952, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Fig.9, caption=Segmentation process visualization, figureFileSmall=r0NXX7KH94eEQaBngwDZ9g==, figureFileBig=1Kf0Atrr4f0UzMBtBsgseQ==, tableContent=null), ArticleFig(id=1218964284005077944, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=图9, caption=分割过程可视化, figureFileSmall=r0NXX7KH94eEQaBngwDZ9g==, figureFileBig=1Kf0Atrr4f0UzMBtBsgseQ==, tableContent=null), ArticleFig(id=1218964284109935550, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Table 1, caption=Classification standard of algae cover grade, figureFileSmall=null, figureFileBig=null, tableContent=
藻类覆盖等级憎水性等级藻类状态描述
a级(轻)HC1～2小面积浅绿色的藻类，藻类分布状态、覆盖程度如图5绿色区域
b级(一般)HC3～4小面积绿色的藻类，藻类分布状态、覆盖程度如图5黄色区域
c级(中)HC5～6大面积绿色的藻类，伞裙表面基本完全覆盖浅绿色的藻类，藻类分布状态、覆盖程度如图5紫色区域
d级(重)HC7伞裙表面完全覆盖深绿色的藻类，藻类分布状态、覆盖程度如图5蓝色区域
), ArticleFig(id=1218964284202210243, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=表1, caption=

藻类覆盖等级分类标准

, figureFileSmall=null, figureFileBig=null, tableContent=
藻类覆盖等级憎水性等级藻类状态描述
a级(轻)HC1～2小面积浅绿色的藻类，藻类分布状态、覆盖程度如图5绿色区域
b级(一般)HC3～4小面积绿色的藻类，藻类分布状态、覆盖程度如图5黄色区域
c级(中)HC5～6大面积绿色的藻类，伞裙表面基本完全覆盖浅绿色的藻类，藻类分布状态、覆盖程度如图5紫色区域
d级(重)HC7伞裙表面完全覆盖深绿色的藻类，藻类分布状态、覆盖程度如图5蓝色区域
), ArticleFig(id=1218964284294484936, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Table 2, caption=Effect of different algal coverage levels, figureFileSmall=null, figureFileBig=null, tableContent=
网络类型类别像素准确率平均值均交并比平均值骰子系数豪斯多夫距离相关体积误差

M-UNet

（本文）

绝缘子本体0.960.920.930.840.9432.860.03
藻类等级10.860.72
藻类等级20.870.78
藻类等级30.880.70
藻类等级40.980.94
Deeplabv3+绝缘子本体0.950.900.920.810.8335.560.04
藻类等级10.790.64
藻类等级20.870.74
藻类等级30.850.68
藻类等级40.960.90
Transfuse绝缘子本体0.950.830.880.740.8342.370.05
藻类等级10.670.54
藻类等级20.760.66
藻类等级30.730.59
藻类等级40.890.79
), ArticleFig(id=1218964284411925454, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=表2, caption=

不同算法藻类覆盖等级的效果

, figureFileSmall=null, figureFileBig=null, tableContent=
网络类型类别像素准确率平均值均交并比平均值骰子系数豪斯多夫距离相关体积误差

M-UNet

（本文）

绝缘子本体0.960.920.930.840.9432.860.03
藻类等级10.860.72
藻类等级20.870.78
藻类等级30.880.70
藻类等级40.980.94
Deeplabv3+绝缘子本体0.950.900.920.810.8335.560.04
藻类等级10.790.64
藻类等级20.870.74
藻类等级30.850.68
藻类等级40.960.90
Transfuse绝缘子本体0.950.830.880.740.8342.370.05
藻类等级10.670.54
藻类等级20.760.66
藻类等级30.730.59
藻类等级40.890.79
), ArticleFig(id=1218964284516783057, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=EN, label=Table 3, caption=Comparison of the effect of M-UNet and U-Net on the classification of algae coverage, figureFileSmall=null, figureFileBig=null, tableContent=
网络类型像素准确率均交并比骰子系数豪斯多夫距离相关体积误差
M-UNet0.920.840.9432.860.03
VGG16+U-Net0.840.740.9141.070.06
U-Net0.650.560.8844.630.09
), ArticleFig(id=1218964284596474835, tenantId=1146029695717560320, journalId=1149653034449285133, articleId=1210613541879541954, language=CN, label=表3, caption=

M-UNet与U-Net藻类覆盖等级分割效果对比

, figureFileSmall=null, figureFileBig=null, tableContent=
网络类型像素准确率均交并比骰子系数豪斯多夫距离相关体积误差
M-UNet0.920.840.9432.860.03
VGG16+U-Net0.840.740.9141.070.06
U-Net0.650.560.8844.630.09
)], attaches=null, journal=Journal(id=1146437660891000859, delFlag=0, nameCn=绝缘材料, nameEn=Insulating Materials, nameHistory1=null, nameHistory2=null, issn=1009-9239, eissn=null, cn=45-1287/TM, coden=null, periodic=0, language=CN, oaType=是, ccby=CC BY-NC-ND, 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=To3JhdgsOj1Q7eh5WnUN1g==, journalPrice=null, startedYear=null, abbrevIsoEn=null, journalRemark=null, publicationField=null, createdTime=1751261638140, updatedTime=1761735740591, createdBy=18614031015, updatedBy=13701087609, firstLetterCn=I, firstLetterEn=I, subjectCode=Natural Sciences, subjectName=Natural Sciences, subjectCodeEn=Natural Sciences, subjectNameEn=null, picCn=To3JhdgsOj1Q7eh5WnUN1g==, picEn=O7zoVkJ9hJoi8iGUSQYTxg==, jcr=null, cjcr=null, exts=[JournalExt(id=1190369230903152828, language=CN, name=绝缘材料, 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=1761735740638, updatedTime=1761735740638, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://jyct.cbpt.cnki.net/EditorDN/index.aspx?t=1, submissionEditorUrl=https://jyct.cbpt.cnki.net/EditorDN/index.aspx?t=3, submissionReviewUrl=https://jyct.cbpt.cnki.net/EditorDN/index.aspx?t=2, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1190369230945095869, language=EN, name=Insulating Materials, 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=1761735740648, updatedTime=1761735740648, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://jyct.cbpt.cnki.net/EditorDN/index.aspx?t=1, submissionEditorUrl=https://jyct.cbpt.cnki.net/EditorDN/index.aspx?t=3, submissionReviewUrl=https://jyct.cbpt.cnki.net/EditorDN/index.aspx?t=2, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1149653034449285133, websiteList=[Website(id=1189939819993166814, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1149653034449285133, 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/jycl/CN, language=CN, createTime=1761633361099, createBy=18614031015, updateTime=1761633401425, updateBy=18614031015, name=绝缘材料-中文, tplId=1146099689490845704, title=绝缘材料, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1189940292275991527, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939819993166814, code=articleTextType, value=kx, createTime=1761633473700, updateTime=1761633473700, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940292250825700, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939819993166814, code=banner, value=null, createTime=1761633473694, updateTime=1761633473694, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940292301157354, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939819993166814, code=grayFlag, value=0, createTime=1761633473706, updateTime=1761633473706, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940292238242787, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939819993166814, code=logo, value=https://castjournals.cast.org.cn/joweb/jycl/CN/file/pic?fileId=MyqZAHzZT6tMetr2hjDKLQ==, createTime=1761633473691, updateTime=1761633473691, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940292322128876, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939819993166814, code=minRunFlag, value=0, createTime=1761633473711, updateTime=1761633473711, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940292267602918, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939819993166814, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/jycl/CN/file/pic, createTime=1761633473698, updateTime=1761633473698, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940292309545963, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939819993166814, code=silenceFlag, value=0, createTime=1761633473708, updateTime=1761633473708, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940292259214309, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939819993166814, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1761633473696, updateTime=1761633473696, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940292284380136, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939819993166814, code=themeColor, value=null, createTime=1761633473702, updateTime=1761633473702, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940292292768745, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939819993166814, code=themeStyle, value=null, createTime=1761633473704, updateTime=1761633473704, creator=18614031015, updator=18614031015)]), Website(id=1189939820060275680, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1149653034449285133, 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/jycl/EN, language=EN, createTime=1761633361115, createBy=18614031015, updateTime=1761633397117, updateBy=18614031015, name=绝缘材料-英文, tplId=1146101810881728533, title=Insulating Materials, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1189940323548722161, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939820060275680, code=articleTextType, value=kx, createTime=1761633481156, updateTime=1761633481156, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940323519362030, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939820060275680, code=banner, value=null, createTime=1761633481149, updateTime=1761633481149, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940323569693684, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939820060275680, code=grayFlag, value=0, createTime=1761633481161, updateTime=1761633481161, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940323506779117, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939820060275680, code=logo, value=https://castjournals.cast.org.cn/joweb/jycl/EN/file/pic?fileId=MyqZAHzZT6tMetr2hjDKLQ==, createTime=1761633481146, updateTime=1761633481146, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940323586470902, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939820060275680, code=minRunFlag, value=0, createTime=1761633481165, updateTime=1761633481165, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940323540333552, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939820060275680, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/jycl/EN/file/pic, createTime=1761633481154, updateTime=1761633481154, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940323578082293, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939820060275680, code=silenceFlag, value=0, createTime=1761633481163, updateTime=1761633481163, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940323531944943, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939820060275680, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1761633481152, updateTime=1761633481152, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940323557110770, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939820060275680, code=themeColor, value=null, createTime=1761633481158, updateTime=1761633481158, creator=18614031015, updator=18614031015), WebsiteProps(id=1189940323565499379, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189939820060275680, code=themeStyle, value=null, createTime=1761633481160, updateTime=1761633481160, creator=18614031015, updator=18614031015)])], journalTitle=绝缘材料, weixinUrl=null, journalUrl=https://www.insulation.org.cn/, iacademicId=null, status=1, seqNo=null, journalTitleEn=Insulating Materials, journalPhotoCn=To3JhdgsOj1Q7eh5WnUN1g==, journalPhotoEn=O7zoVkJ9hJoi8iGUSQYTxg==, journalFirstLetter=I, 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/jycl/CN/10.16790/j.cnki.1009-9239.im.2024.11.017, detailUrlEn=https://castjournals.cast.org.cn/joweb/jycl/EN/10.16790/j.cnki.1009-9239.im.2024.11.017, pdfUrlCn=https://castjournals.cast.org.cn/joweb/jycl/CN/PDF/10.16790/j.cnki.1009-9239.im.2024.11.017, pdfUrlEn=https://castjournals.cast.org.cn/joweb/jycl/EN/PDF/10.16790/j.cnki.1009-9239.im.2024.11.017, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
融合多尺度卷积注意力机制的绝缘子表面藻类污秽语义分割和识别算法研究
收藏切换
PDF下载
杨世芳 1 , 臧欣怡 1 , 朱明曦 3 , 刘云鹏 1 , 石超君 2 , 贾志东 4
绝缘材料 | 测试与分析 2024,57(11): 135-143
收起
收藏切换
绝缘材料 | 测试与分析 2024, 57(11): 135-143
融合多尺度卷积注意力机制的绝缘子表面藻类污秽语义分割和识别算法研究
全屏
杨世芳1, 臧欣怡1, 朱明曦3, 刘云鹏1, 石超君2, 贾志东4
作者信息
  • 1华北电力大学，电力工程系，河北 保定 071000
  • 2华北电力大学，电子与通信工程系，河北 保定 071000
  • 3国网陕西省电力公司电力科学研究院，陕西 西安 710100
  • 4清华大学深圳国际研究生院，广东 深圳 518055

    • 杨世芳（1994-），女（汉族），河北保定人，讲师，主要从事高电压绝缘生物染污问题、退役绝缘材料无害化处理等方面的研究。

Research on semantic segmentation and recognition algorithm integrating multi-scale convolutional attention mechanism for algae pollution on insulator surface
Shifang YANG1, Xinyi ZANG1, Mingxi ZHU3, Yunpeng LIU1, Chaojun SHI2, Zhidong JIA4
Affiliations
  • 1Department of Electric Power Engineering, North China Electric Power University, Baoding 071000, China
  • 2Department of Electronic and Communication Engineering, North China Electric Power University, Baoding 071000, China
  • 3State Grid Shaanxi Electric Power Research Institute, Xi′an 710100, China
  • 4Tsinghua Shenzhen International Graduate School, Shenzhen 518055, China
出版时间: 2024-11-20 doi: 10.16790/j.cnki.1009-9239.im.2024.11.017
文章导航
收藏切换
摘要
收起

藻类是一种特殊生物污秽，其在电气设备绝缘子表面附着会使绝缘子耐污闪能力大幅下降，给电网安全稳定运行带来威胁。本文提出了一种融合多尺度卷积注意力机制的绝缘藻类语义分割算法。首先在基础U-Net网络模型上构建了一种用于绝缘藻类等级语义分割的模型，并采用VGG16作为主干特征提取网络。模型采用U型结构，左侧为VGG16及主干特征提取部分，能够有效提取5个特征层信息；右侧为加强特征提取部分。注意力模块选取CBAM模块，并在其基础上引入多尺度卷积；随后在U-Net网络的编码器和解码器进行上采样与下采样之前均加入CBAM卷积注意力模块。最后将本文模型在自行构建的藻类覆盖绝缘子图像数据集上与Deeplabv3+与Transfuse网络进行对比。结果表明：相比于基础U-Net模型，本文模型mIoU值提高了0.28，mPA值提高了0.27，Dice系数提高了0.06，豪斯多夫距离降低了11.77，RVE值降低了0.06。分割过程可视化结果展示了本文模型能更好地关注藻类覆盖区域，且对藻类覆盖区域边界的定位更准确，有效减小了分割误差。

绝缘藻类  /  图象处理  /  语义分割  /  注意力机制

Algae is a special kind of biofouling, its attachment on the insulator surface of electrical equipment will significantly reduce the fouling flash resistance of insulator , which poses a threat to the safe and stable operation of power grids. In this paper, a semantic segmentation algorithm of insulating algae integrated multi-scale convolution attention mechanism was proposed. Firstly, a model for insulating algae class semantic segmentation was constructed on the basic U-Net network model, and VGG16 was used as the backbone feature extraction network. The model adopts a U-shaped structure, and the left side is the feature extraction part of the VGG16 backbone, which can effectively extract the informations of five feature layers. The right side was the enhanced feature extraction part. CBAM module was selected for attention module, and the multi-scale convolution was introduced based on CBAM module. Then the CABM convolutional attention module was added to the encoder and the decoder of U-Net network before up-sampling and down-sampling. Finally, the model was compared with Deeplabv3+ and Transfuse network on the self-constructed algae-covered insulator image dataset. The results show that compared with the basic U-Net model, the mIoU value of this model improves by 0.28, mPA value improves by 0.27, Dice coefficient improves by 0.06, Hausdorff distance reduces by 11.77, and the RVE value reduces by 0.06. The visualization results of the segmentation process demonstrates that the model in this paper can pay more attention to the algal coverage region, and locate the boundary of the algal coverage region more accurately, which reduces the segmentation error effectively.

insulating algae  /  image processing  /  semantic segmentation  /  attention mechanisms
杨世芳, 臧欣怡, 朱明曦, 刘云鹏, 石超君, 贾志东. 融合多尺度卷积注意力机制的绝缘子表面藻类污秽语义分割和识别算法研究. 绝缘材料, 2024 , 57 (11) : 135 -143 . DOI: 10.16790/j.cnki.1009-9239.im.2024.11.017
Shifang YANG, Xinyi ZANG, Mingxi ZHU, Yunpeng LIU, Chaojun SHI, Zhidong JIA. Research on semantic segmentation and recognition algorithm integrating multi-scale convolutional attention mechanism for algae pollution on insulator surface[J]. Insulating Materials, 2024 , 57 (11) : 135 -143 . DOI: 10.16790/j.cnki.1009-9239.im.2024.11.017
正文
收起
0 引言
收起
在高压输电系统中，绝缘子是非常重要的外绝缘设备，具有电气绝缘和机械支撑的双重作用。随着电网规模的日益扩大和环境污染的日益严重，绝缘子污闪已成为影响电力系统安全稳定运行的最大威胁[1]。
导致绝缘子发生污闪的原因有很多：如绝缘子自身的耐污能力低、绝缘子运行环境污染较大、绝缘子所处地区气候恶劣以及绝缘子运行维护不到位等[2-3]。近年来，随着我国不断加强生态文明建设，大力保护生态环境，绿植附着率明显提高，高压输电线路周围的生物也逐渐增多。在我国南方潮湿地区，藻类生物污秽大量出现在复合绝缘子和RTV涂层绝缘子表面[3-6]。
藻类通过吸收水分和湿气来维持新陈代谢，因此表现出很强的亲水性[7-9]。藻类附着在硅橡胶绝缘材料表面时，会使硅橡胶绝缘材料的憎水性能降低甚至丧失，导致其耐污闪能力大幅下降[10-11]，给输变电设备的安全运行带来威胁。
目前关于绝缘子表面生长藻类的研究处于起步阶段，研究内容涉及绝缘子表面藻类的种类鉴别、藻类对硅橡胶表面憎水性能的影响、藻类对硅橡胶的降解作用以及藻类对绝缘子闪络特性的影响。相关研究尚未涉及到绝缘子表面藻类生长量的测量和生长程度的分级，也未涉及图像处理对绝缘子表面藻类的识别[12-13]。
传统绝缘子缺陷检测一般通过人工巡检的方式进行，架空输电线路数量的增加使巡检规模更加庞大、巡检环境更加复杂，放大了传统绝缘子缺陷检测方法人力成本高、检测效率低的不足[14]。无人机（unmanned aerial vehicle，UAV）等新型巡线方式可依靠深度学习目标检测算法识别架空输电线路绝缘子的藻类覆盖情况以及藻类覆盖率[15]。目前，电厂已经使用无人机巡检并辅以深度学习目标检测算法，用于输电线路缺陷识别检测以及电晕放电检测[35]等。现亟需一种比原始色度分割法（RGB分割法）更为简单便利的方法来标记绝缘体表面藻类的覆盖程度。这项任务可通过深度学习算法应用于绝缘子表面藻类覆盖等级分割领域来实现。
深度学习算法具有良好的挖掘和刻画问题复杂内部特征的能力以及强大的学习能力[18]，因此能够直接从原始信息中获得问题深度特征。利用卷积神经网络自动提取图像深度特征[19]，展示出了比传统绝缘子缺陷检测算法更好的能力。深度卷积神经网络架构（VGG）使用3×3卷积核，并且使用了大小相同的最大池化，在保证感受野相同的情况下，网络的深度和特征提取的效果都得到了提高，通过加深网络结构提高了性能。深度卷积神经网络架构U-Net语义分割模型最早是用于解决医学图像分割的问题，与其他的深度学习神经网络模型相比，该模型不仅支持少量的数据训练模型，而且对图像的分割准确率更高，分割速度更快[20-21]。U-Net算法网络通过增加特征提取网络深度来改善像素分割效果，从而实现快速高效地判断绝缘子藻类覆盖情况以及藻类覆盖度等级划分。
卷积块注意力（CBAM）模块是一种用于增强卷积神经网络性能的注意力机制[22-23]，它能通过自适应学习特征图之中不同空间位置和通道维度之间的关系来提取更加有区分度的特征[24-26]。CBAM模块先采用注意力机制对特征图进行通道维度上的特征提取，再对空间位置维度上的注意力机制进行融合，提取不同维度特征之间的相关信息。多尺度的特征融合提高了卷积神经网络模型的表达能力和特征鉴别能力[27-29]。
鉴于U-Net语义分割模型在生物医学图像领域上成功的应用，本文提出一种基于U-Net语义分割模型并结合VGG16引入CBAM卷积注意机制的聚合网络模型，首次将U-Net算法网络引入藻类覆盖等级分割领域，对基础的U-Net算法网络加以改进，使用VGG16作为主干特征提取网络，添加通道注意力机制及空间注意力机制，将CBAM模块融合多尺度卷积，并加入U-Net算法网络，提出了M-UNet模型；并自行构建藻类覆盖绝缘子数据集，以验证改进后的M-UNet模型分割藻类覆盖等级的可行性，并在数据集上使用Deeplabv3+、Transfuse网络进行实验，使结果与M-UNet模型进行对比。M-UNet模型的应用为分割藻类覆盖等级，及时安排合适的藻类污染防治具体措施提供了技术支撑。
1 绝缘藻类语义分割算法
收起
1.1 M-UNet算法网络结构
为实现对覆藻绝缘子图像的绝缘子区域和藻类覆盖区域信息的提取以及藻类覆盖等级的分割，本文以加强提取藻类覆盖特征为核心，提出了一种融合多尺度卷积注意机制的绝缘藻类语义分割算法。
绝缘藻类语义分割算法主干特征提取网络为VGG16，它是一种卷积神经网络，VGG16网络结构如图1所示。
使用VGG16作为主干特征提取网络，需要构建卷积层和最大池化层。输入图像大小为512×512×3，基础的VGG结构因全连接层参数较多引发高内存、运行时间长的问题，不适合应用于图像分割领域，故去掉全连接层以加快运算速度和防止网络出现过拟合问题，在保证具有相同感受野的条件下提升网络深度，并补充边缘信息，从而在一定程度上提升神经网络效果。当输入图像大小为512×512×3时，VGG16网络特征融合过程如图2所示。
（1）卷积层，为更好地捕捉图像变化的细节，获得更好的识别效果，减少参数数量，经过大小为3×3步长（stride）为1，填充（padding）为1的卷积操作，可不改变特征矩阵的高和宽、防止梯度消失、减少过拟合、提高网络的训练速度，在卷积层后加入激活函数（ReLU）。
（2）最大池化层，通过将最大池化下采样的特征矩阵高和宽均缩减为原来的1/2，从而得到特征图。最大池化层可以减小卷积层参数误差引起的平均估计偏差，更容易捕捉图像和渐变中的变化，有利于保留纹理等详细信息。
ReLU函数是一种分段线性函数，本文模型采用ReLU函数作为激活函数。ReLU函数在负半区的导数为0，即神经元激活值进入负半区，梯度变为0（单侧抑制），拥有较快的计算速度，其公式如式（1）所示。
fx=max0, x,   x≥0       0,             x<0  
1.2 CBAM模块
此外，为了突出藻类分布区域特征信息以及加强不同等级藻类覆盖特征信息的提取，投入更大的权重，突出和加强有用特征，抑制和忽略无关特征。本文在U-Net算法网络中添加CBAM模块，算法结构如图3所示。通道注意力模块通过藻类特征通道间的关系来生成通道注意力图。为了更好地计算通道注意力，CBAM模块压缩了输入特征映射的空间维度，在空间信息聚合中采用平均池化法，能更好地理解通道之间的全局依赖关系。空间注意力模块用于学习空间之间的相关关系，利用特征之间的通道关系生成通道注意图，作为本模块的输入特征图。空间注意模块关注局部信息，首先对信息进行合并过滤，然后对过滤后的信息进行卷积并提取重要信息。通过将通道注意力模块与空间注意力模块结合，CBAM模块能够同时学习通道和空间注意力，并将其应用于不同层的神经网络模型中，从而帮助U-Net算法网络更好地理解不同等级藻类特征的重要性，并提高网络的性能和对抗扰动的鲁棒性。CBAM模块将通道注意力子模块放在空间注意子模块之前使得其获得更好的特征提取能力[30]，因此本文使用先通道注意力后空间注意力的顺序。
绝缘子表面覆盖不同等级的藻类之间颜色、纹理的差别并不明显，导致网络分割藻类覆盖等级准确率较低。注意力机制能快速扫描全局图像，获取图像焦点的同时过滤掉无用信息[31]。为提高CBAM注意力机制提取目标信息能力，将多尺度卷积引入CBAM模块中，出于算法的便利性也可减少网络参数，只使用3×3的卷积，同时结构中保留一个未操作的路径可用于增加网络性能。原本的CBAM模块只对原特征图进行最大池化和平均池化操作，增加了使用卷积、拼接的多尺度卷积运算来生成新的特征图。引入多尺度卷积能够提升运算所得的通道注意权重和空间注意权重的感受野，强调藻类颜色、分布、边界特征信息，同时过滤掉其他冗余灰尘、污染等信息。
为了提高U-Net算法网络对藻类覆盖绝缘子的分割精度，本文通过编码器和解码器进行上采样和下采样前均添加了CBAM模块。添加了CBAM模块后的M-UNet算法网络模型如图4所示。
2 绝缘藻类语义分割算法网络性能测试
收起
为证明本文所提覆藻绝缘子藻类覆盖等级分割算法的有效性，进行了实验验证。实验环境为：CPU：Intel Core i7-7700HQ、8G、NVIDIA GeForce GTX 1050Ti；使用windows10操作系统中的Phycharm 2021.1.2IDE、Python3.8，基于Pytorch1.7.1深度学习框架。
2.1 构建藻类覆盖绝缘子数据集
（1）数据集准备
准备不同藻类污染程度的绝缘子，其中包含110 kV支柱绝缘子、10 kV悬式电瓷绝缘子、10 kV悬式玻璃绝缘子和10 kV复合绝缘子。拍摄前先把数码相机预设白平衡，将绝缘子竖直放置，使绝缘子伞裙处于水平面内，手持数码相机，保持数码相机与所要拍摄伞裙之间的距离为30～50 cm，同一薄膜污染程度绝缘子从垂直上方拍摄1张照片，拍摄视线与伞裙表面夹角大于60°，使拍摄视线与伞裙表面尽量垂直，拍摄绝缘子4张照片，共拍摄5张照片。光源方向同拍摄方向保持一致，确定绝缘子直视部位无影子。
藻类覆盖绝缘子数据采集共计拍摄绝缘子图像1 140张，将图像经过顺时针旋转5°、逆时针旋转5°以及左右镜像等方法[32-33]，将绝缘子图像扩充到5 340张作为藻类分割绝缘子数据集。
（2）标签制作
将绝缘子单独标注为1个标签，根据《绝缘子藻类现场覆盖度等级划分和防治技术导则》，将藻类覆盖等级分为a级（轻）～d级（重）共4个等级，表1为藻类覆盖的4个分类等级标准。
使用labelme软件对藻类覆盖绝缘子图像进行标注后，选取清晰稳定的500张照片作为测试集，将原藻类覆盖绝缘子图像与标签文件合并放入数据集文件夹为实验做准备。
2.2 实验步骤
使用M-UNet模型以及基础的U-Net算法网络、Deeplabv3+、Transfuse网络进行绝缘子与藻类覆盖区域等级分割，整个过程可分为4个步骤：绝缘子图像拍摄、图像分割、数据处理和结果输出，过程如图6所示。第一，制作数据集，从采集好的图片中进行挑选，去除模糊、杂乱的图像后，挑选出5 000张作为数据总集，按9∶1的比例选出训练集与预测集，并对数据集进行标定。第二，在M-UNet主干结构下开始对训练集中的5 000张图像进行训练。本实验初始学习率设置为0.000 5，训练周期设置为1 000，批大小设置为2。采用ReLU函数为激活函数，Diss Loss函数为损失函数，并采用Adam优化算法优化网络参数。在训练过程中，观察损失函数收敛情况，同时调节网络相关参数来优化结果。第三，通过不断迭代训练，获得最优的模型迭代结果，将最优的输出结果进行保存，对测试集图像进行测试。最后，在最优的输出结果出现后，输出中间特征图，并将数据集在Deeplabv3+、Transfuse网络上进行测试并输出中间特征图，将分割过程可视化，并对比。
2.3 评价指标
（1）像素准确率
像素准确率表示预测类别正确的像素数占总像素数的比例，其公式如式（2）所示。
PA=TP+TNTP+TN+FN
式（2）中：TP表示正确预测绝缘子以及藻类覆盖部分及等级的区域；FN表示预测错误的绝缘子以及藻类覆盖区域；FP表示预测正确的绝缘子以及藻类覆盖区域。
（2）均交并比
均交并比公式如式（3）所示。
MIOU=1k+1∑i=0kTPTP+TN+FN
式（3）中：k表示藻类等级以及绝缘子分类的类别；i表示真实值。
（3）骰子系数
骰子系数（Dice）是一种集合相似度度量函数，用于计算两个样本的相似度，取值范围为[0,1]，其公式如式（4）所示。
Dice=2X⋂YX+Y
式（4）中：X，Y分别为预测结果图和真实标签图；∩为两个像素集对应点的与运算；丨丨为像素数量的代数求和运算。
（4）豪斯多夫距离
豪斯多夫距离（hausdorff distance，HD），越小代表预测边界分割误差越小、质量越好。给定欧氏空间中的两点集A={a1，a2，a3，⋯，an}和B={b1，b2，b3，⋯，bm}，则豪斯多夫距离如式（5）所示。
HA, B=maxhA, B, hB, A 
式（5）中：H(A,B)=maxa∈Aminb∈B||a-b||，A和B表示两个不相交的集合，HD根据式（5）测量2个等值线之间的距离。
3 绝缘藻类语义分割算法网络性能分析
收起
3.1 损失函数
在M-UNet算法网络上训练本文所构建的数据集，得到训练过程的损失函数图。在训练中迭代300轮，其中数据集中用于训练、验证的数据比例为9∶1，并以不同的Adam进行迭代优化。在训练模型的同时，监测训练数据和验证数据的损失，改进后的M-UNet、Deeplabv3+、Transfuse 3种模型的损失函数曲线如图7所示。
从图7可以看出，改进后的M-UNet模型损失值经过细微的波动后，在230轮左右趋于平稳。随着训练次数增加，训练集的损失函数逐渐减小，收敛速度较快并且稳定收敛在0.07附近，Deeplabv3+网络损失函数收敛在0.06附近，Transfuse网络损失函数收敛在0.13附近。M-UNet模型的损失函数曲线在模型训练初期就表现出明显的优势，其损失值下降速度最快，且在模型训练结束后得到了最小的损失值。且改进后的M-UNet模型的损失函数曲线振荡幅度小，模型比较稳定，因此在绝缘子藻类覆盖等级分割上表现出极大的优势。
3.2 分割效果
实验中将藻类覆盖绝缘子等级分割成5个部分，分别为绝缘子本体以及藻类污染程度1、2、3、4级，将5个部分经过M-UNet模型运算，结果如表2所示。
像素准确率与均交并比值越大，说明藻类等级分割的越精准。从表2可以看出，M-UNet模型的平均像素准确率与平均交并比分别为0.92和0.84，远高于Deeplabv3+模型以及Transfuse模型。M-UNet网络通过编码器和解码器进行上采样和下采样之前均添加了卷积注意力机制，提高了卷积层提取特征的能力。并将藻类覆盖绝缘子图像每个像素点精确映射到高维分布上，CBAM的加入提升了网络的分割精度。
Dice系数用于评估有限样本集之间的相似性，Dice系数越大，相似度越高。M-UNet模型在绝缘子藻类覆盖等级分割结果的Dice系数比Deeplabv3+模型提高了0.11，比Transfuse模型提高了0.11，说明了M-UNet网络对藻类覆盖等级分割的有效性。改进后的U-Net模型Dice系数相比基础U-Net模型提高了0.06，表明注意力机制的添加进一步提高了网络对藻类覆盖等级分割的有效性。
豪斯多夫距离作为一种相似性度量，用于衡量有限样本集之间的相似性，两种样本相似性越高，HD数值越小。M-UNet模型的HD比Deeplabv3+模型降低了2.70，相比Transfuse模型降低了9.51。通过Dice系数与豪斯多夫距离的变化，表明改进后的M-UNet模型对绝缘子藻类覆盖区域轮廓的识别相似性高。
相关体积误差用于衡量藻类等级分割结果与藻类等级真实分割结果之间的体积差异，其值越小，表示分割结果与真实分割的体积越接近。改进后的M-UNet网络相关体积误差结果为0.03，较Deeplabv3+与Transfuse模型分别降低了0.01、0.02。进一步证明了改进后的M-UNet模型对绝缘子藻类覆盖等级分割的高准确率。
图8为3种网络下绝缘子藻类覆盖等级的分割结果。由图8中5幅原始图像可以看出，每幅图像的藻类覆盖程度、大小、颜色都有很明显的变化，且在实际场地中绝缘子表面不止有藻类还有灰土以及其他污染物，导致绝缘子表面不止有绿色，还有灰白色以及黑色等其他颜色的污秽，为精准分割增添了很大的难度。
从图8还可以看出，改进后的M-UNet网络可以更好地识别藻类覆盖区域，且对于不同等级藻类覆盖的分割效果更好。在藻类轻度污染情况下，Deeplabv3+和Transfuse会将藻类覆盖区域识别为绝缘子，而M-UNet网络可完美地识别藻类覆盖区域，且更为清晰、准确地划分覆藻区域边界。在藻类污染程度复杂度比较高的区域，与Deeplab3+和Transfuse网络相比，M-UNet网络在识别相邻污染程度的藻类覆盖区域方面展示出了显著优势。VGG16主干特征提取网络的加入缩短了网络训练时间；而注意力机制的加入不仅突出了不同藻类覆盖污染程度的颜色、分布及边界特征信息，还能有效过滤掉其余污染物的视觉干扰，如灰尘和其他污染等，使网络能更好地提取藻类污染区域。综上，VGG16主干特征提取网络与注意力机制的引入有效提高了网络藻类覆盖等效划分的准确性，对藻类覆盖区域边界的识别更为准确，从而达到更好的覆藻绝缘子藻类覆盖等级划分效果。
3.3 消融实验
本文为验证M-UNet网络对绝缘子藻类覆盖等级分割的有效性，对M-UNet与U-Net的藻类覆盖等级分割效果进行对比，结果如表3所示。从表3可以看出，VGG16主干特征提取网络的添加提高了网络识别的准确性。但仍未能达到藻类覆盖等级分割的预期目标。注意力机制模块的添加显著提高了网络对不同藻类覆盖等级区域分割的能力。Dice系数与HD值的提高证实改进后的网络对不同藻类覆盖等级以及藻类覆盖等级边界的识别更为准确。相关体积误差的减小进一步证明了网络分割精度的提高，证明了U-Net网络中VGG16主干特征提取网络与注意力机制的添加可以提高模型藻类覆盖等级的分割性能。
3.4 分割过程可视化
图9为3种模型的图像分割过程可视化结果。
从图9可以看出，VGG16与CBAM的添加使得M-UNet网络在识别绝缘子表面不同藻类污染等级任务上性能得到提升。在编码器与解码器均添加了注意力机制的M-UNet网络在识别藻类覆盖区域边界时相比于传统的U-Net网络能够较为准确地识别其覆盖区域，同时能更清晰地识别其边界，有效减小了其分割误差。
4 结论
收起
本文收集了包含5 340张藻类覆盖绝缘子图像及其响应分割标签的数据集，并结合VGG16以及卷积注意力模块测试了两种基本图像分割网络模型，得出以下结论：
（1）M-UNet网络实现了平均交并比为0.84、平均像素准确率为0.92的高精度分割结果，完成了对绝缘子以及藻类覆盖等级的高精准分割。
（2）使用现存方法进行图像识别时需提前去除干扰识别的杂点与噪声点，而U-Net模型可去除全部干扰点，M-UNet网络的分割效果亦不会受到外界光照、噪点等因素的影响，因此能够将绝缘子以及不同等级的藻类覆盖区域进行精准分割。
（3）M-UNet模型对藻类等级的分割与Deeplabv3+、Transfuse模型相比，对于光线、拍摄角度、风等外界环境的干扰有较强的鲁棒性与泛化能力。
基金
收起
  • 国家自然科学基金资助项目(52207154)
  • 河北省自然科学基金资助项目(E2023502087)
  • 中央高校基本科研业务费专项资金资助项目(2023MS104)
文献
收起
参考文献 引证文献
排序方式：
[1]
宿志一,李庆峰.我国电网防污闪措施的回顾和总结[J].电网技术,2010,34(12):124-130.
SUZ Y, LIQ F. Historical review and summary on measures against pollution flashover occurred in power grids in China[J]. Power System Technology,2010,34(12):124-130.
[2]
AGARWALM, GUPTAS K, BISWASK K. Development of a compressed FCN architecture for semantic segmentation using particle swarm optimization[J]. Neural computing & applications,2023,35(16):11833-11846.
[3]
FERNANDOM, GUBANSKIS. Closure to discussion of performance of nonceramic insulators under tropical field conditions [J]. IEEE Transactions on Power Delivery,2000,15(4):335-360.
[4]
FLEMMINGlH C. Relevance of biofilms for the biodeterioration of surfaces of polymeric materials[J]. Polymer Degradation & Stability,1998,59(1):309-315.
[5]
ROJASH E, ROJASH D. Performance evaluation of ceramic and polymeric electrical distribution insulators under fungus biocontamination[J]. Información tecnológica,2017,28(2):3-14.
[6]
GORURR S, BERNSTEINB S. Field and laboratory aging of polymeric distribution cable terminations. I. field agin [J]. IEEE Transactions on Power Delivery,1998,13(2):316-322.
[7]
COOWARF, HOOLEP R P. Discharge characteristics of alga-contaminated string insulators in the tropics[J]. Electric Power Systems Research,1988,15(3):215-218.
[8]
GUH, MINZ, WENW, et al. Improved U-Net based insulator image segmentation method based on attention mechanism[J]. Energy Reports,2021,7(9):210-217.
[9]
NAITOK, MATSUNKAR, SAKANISHIK. Investigation of the insulation performance of the insulator covered with lichen[J]. IEEE Transactions on Power Delivery,1990,5(3):1634-1640.
[10]
HEH, HUZ, WANGB, et al. A contactless zero-value insulators detection method based on infrared images matching[J]. IEEE Access,2020,8(5):133882-133889.
[11]
ROBERTD M, ROBERTD H, JOEM K, et al. A study of biological contaminants on high voltage porcelain insulators[J]. Electric Power Systems Research,1996,42(1):35-39.
[12]
YIL, LiZ, LIJ, et al. Fault diagnosis of rotating machinery based on combination of deep belief network and one-dimensional convolutional neural network[J]. IEEE Access,2019,7(8):165710-165723.
[13]
DA Z, SHUAIL C. Insulator contamination grade recognition using the deep learning of color information of images[J]. Energies,2021,14(20):6662-6668.
[14]
李佐胜,李文利,姚建刚,等.应用绝缘子红外热像处理的现场污秽等级检测方法 [J].中国电机工程学报,2010,30(4):132-138.
LIZ S, LIW L, YAOJ G, et al. On-site detection of pollution level of insulators based on infrared-thermal-image processing[J]. Proceedings of the CSEE,2010,30(4):132-138.
[15]
WANGZ, AFGANM S, GUW, et al. Recent advances in laser-induced breakdown spectroscopy quantification: from fundamental understanding to data processing[J]. TrAC Trends in Analytical Chemistry,2021,37(8):116-128.
[16]
GUOH L, QIANW M. Strip steel surface defect detecting method combined with a multi-layer attention mechanism network[J]. Measurement Science and Technology,2023,34(5):44-48.
[17]
RONNEBERGERO, FISCHERP, BROXT. U-Net: convolutional net-works for biomedical image segmentation[C]//Proceedings of the 18th International Conference on Medical Image Computing and Computer-Assisted Intervention. Munich, Germany:Springer,2015.
[18]
刘航,汪西莉.基于注意力机制的遥感图像分割模型[J].激光与光电子学进展,2020,57(4):170-180.
LIUH, WANGX L. Remote sensing image segmentation model based on attention mechanism[J]. Laser and Optoelectronics Progress,2020,57(4):170-180
[19]
EVANS, JONATHANL, TREVORD. Fully Convolutional Networks for Semantic Segmentation[J]. IEEE transactions on pattern analysis and machine intelligence,2017,39(4):640-651.
[20]
ZIFEIL, WENZ Y, RURUG, et al. Transattention U-Net for semantic segmentation of poppy[J]. Electronics,2023,12(3):487-497.
[21]
QINX, ZHANGZ, HUANC, et al. U2-Net: going deeper with nested U-structure for salient object detection[J]. Pattern Recognition,2020,106(5):665-675.
[22]
JINGS, JIE X, ALLEN. Z A,et al. Automatic pixel-level pavement sealed crack detection using multi-fusion U-Net network[J]. Measurement,2023,208(6):55-62.
[23]
QIZ, YINGL, YIZ, et al. A comparative study of attention mechanism based deep learning methods for bladder tumor segmentation[J]. International Journal of Medical Informatics,2023, 171(5):104984-104994.
[24]
肖国庆,李雪琪,陈玥丹,等.大规模图神经网络研究综述[J].计算机学报,2024,47(1):148-171.
XIAOG Q, LIX Q, CHENY D. A survey of large-scale graph neural networks[J]. Chinese Journal of Computers,2024,47(1):148-171.
[25]
HANG J, HEM, ZHAOF, et al. Insulator detection and damage identification based on improved lightweight YOLOv4 network[J]. Energy Reports,2021,7(z7):187-197.
[26]
WANGJ, ZHOUY, YANGY. Rapid 3D measurement technique for colorful objects employing RGB color light projection[J]. Applied Optics,2020,59(7):1907-1915.
[27]
WOO S, PARKJ, LEEJ Y, et al. CBAM: convolutional block attention module[C]//Proceedings of the 15th European Conference on Computer Vision (ECCV). Munich, Germany:Springer,2018.
[28]
WANGJ H, ZHOUY G, YANGY X. Three-dimensional shape detection for non uniform reflective objects: combination of color light projection and camera's exposure adjustment[J]. IEEE Sensors Journal,2020,4(9):1-9.
[29]
OKTAYO, SCHLEMPERJ, FOLGOCL L, et al. Attention U-net: learning where to look for the pancreas[J]. Computer Vision and Pattern Recognition,2018,33(4):222-231.
[30]
KAI W, ZHONGZ, CHAOZ C, et al. Intelligent whistling system of rail train based on YOLOv4 and U-net[J]. Applied Sciences,2023,13(3):1685-1695.
[31]
BUSLARVA, IGLOVIKOVV I, KHVEDCHENYAE, et al. Albumentations: fast and flexible image augmentations[J]. Information,2020,11(2):125-131.
[32]
YANB, CHENQ, YER, et al. Insulator detection and recognition of explosion based on convolutional neural networks[J]. International Journal of Wavelets, Multiresolution and Information Processing,2019,17(2):22-28.
[33]
YAOX, LEIL W, ZHENGZ, et al. Consistency-based semi-supervised learning framework for power line insulators detection[J]. Journal of Physics: Conference Series,2022,22(1):44-51.
[34]
杨秋玉,郑小刚,李建兴,等.基于图像处理的复合绝缘子憎水性智能识别方法[J].绝缘材料,2022,55(9):100-106.
YANGQ Y, ZHENGX G, LIJ X, et al. Intelligent recognition method of composite insulator hydrophnbicity based on image processing[J]. Insulating Materials,2022,55(9):100-106.
[35]
范路,陆云才,陶风波,等.人工智能在局部放电检测中的应用(二):模式识别与状态评估[J].绝缘材料,2021,54(7):10-24.
FANL, LUY C, TAOF B, et. al. Application of artificial intelligence in partial discharge detection part Ⅱ: pattern recognition and condition assessment[J]. Insulating Materials,2021,54(7):10-24.
2024年第57卷第11期
PDF下载
126
56
引用本文
BibTeX
文章信息
doi: 10.16790/j.cnki.1009-9239.im.2024.11.017
  • 接收时间：2024-04-04
  • 首发时间：2025-12-24
  • 出版时间：2024-11-20
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期：2024-04-04
  • 修回日期：2024-06-21
基金
国家自然科学基金资助项目(52207154)
河北省自然科学基金资助项目(E2023502087)
中央高校基本科研业务费专项资金资助项目(2023MS104)
作者信息
    1华北电力大学，电力工程系，河北 保定 071000
    2华北电力大学，电子与通信工程系，河北 保定 071000
    3国网陕西省电力公司电力科学研究院，陕西 西安 710100
    4清华大学深圳国际研究生院，广东 深圳 518055
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/jycl/CN/10.16790/j.cnki.1009-9239.im.2024.11.017
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
科
Family		 属数
Number of
genus		 种数
Number of
species		 占总种数比例
Percentage of
total species (%)		 属
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
相关链接
论文
最新文章
热读文章
热点专题
内参
蓝皮书
产业发展报告
传播力报告
期刊分级目录
资讯
学术新闻
行业新闻
学术会议
行业会议
关于
关于我们
联系我们
北京市海淀区学院南路86号
100081
010-62199257
qkjq@cast.org.cn

中国科学技术协会版权所有 京ICP 备10216604号-4 海淀分局备案 1101084647
科技导报社主办 中国科协信息中心技术支持