毛万登,刘善峰,张琦,鲍华
MAO Wandeng,LIU Shanfeng,ZHANG Qi,BAO Hua

• 1:国网河南省电力公司
• 2:安徽大学人工智能学院

摘要(Abstract)：

现有的变电站设备缺陷检测算法存在复杂场景下难以完整检测不同尺度目标、检测结果边界模糊等问题。为此，提出一种基于渐进式特征优化与融合策略的变电站设备缺陷检测方法，通过加强缺陷主体部分特征的学习，优化缺陷主体信息并弱化背景信息，将缺陷主体从复杂背景中分离。首先，引入注意力机制，从空间和通道两方面自适应学习并优化复杂背景中的缺陷特征。其次，采用层次化特征交互策略整合高级语义信息，引导后续各层特征的有效融合。接着，利用渐进式空洞融合模块有效聚合多尺度特征。最后，在构建的SEAD（变电站设备缺陷数据集）上进行对比实验，结果显示，所提方法在SEAD上MAE（平均绝对误差）为0.007 0,F-measure(F度量)为0.962 8，相较于其他方法具有显著的优势。
Current substation equipment anomaly detection methods struggle with incomplete multi-scale target detection and boundary ambiguity in complex scenarios. To address these limitations, this study proposes an anomaly detection method by employing a progressive feature optimization and fusion strategy. This method enhances the learning of features related to anomaly subjects and optimizes the information of anomaly subjects while reducing the influence of background information, effectively separating anomaly subjects from complex backgrounds. Firstly, an attention mechanism is introduced to adaptively learn and optimize anomaly features in complex backgrounds from both spatial and channel dimensions. Secondly, a hierarchical feature interaction strategy is employed to integrate high-level semantic information to guide the effective fusion of features across subsequent layers. Finally, a progressive dilated fusion module is employed to effectively aggregate multi-scale features. Comparative experiments conducted on the newly developed substation equipment anomaly dataset(SEAD) demonstrate the superior performance of the method, achieving a mean absolute error(MAE) of 0.007 0 and F-measure, a similarity metric, of 0.962 8.

关键词(KeyWords)： 缺陷检测;显著性目标检测;多尺度信息融合;渐进式融合
anomaly detection;salient object detection;multi-scale information fusion;progressive fusion

Abstract:

Keywords:

基金项目(Foundation): 安徽省自然科学基金面上项目(1908085MF217);; 安徽省教育厅重点项目(2023AH050085)

作者(Author): 毛万登,刘善峰,张琦,鲍华
MAO Wandeng,LIU Shanfeng,ZHANG Qi,BAO Hua

DOI: 10.19585/j.zjdl.202508012

参考文献(References)：

• [1] WANG J D,CHENG Z,MENG F Q.A review of power equipment defect detection based on deep learning[C]//2023 IEEE 12th International Conference on Communication Systems and Network Technologies(CSNT),April8-9,2023,Bhopal,India.IEEE,2023:120-125.
• [2]韩睿，戴哲仁，蒋鹏，等.基于改进YOLOv8的电力场景通用缺陷检测模型[J].浙江电力，2024,43(4):113-120.HAN Rui,DAI Zheren,JIANG Peng,et al.A general defect detection model for power scenarios using the improved YOLOv8[J]. Zhejiang Electric Power,2024,43(4):113-120.
• [3]季世超，曲星合，宋庆彬，等.基于改进YOLOv5的轻量化绝缘子缺陷检测算法[J].浙江电力，2023,42(12):64-72.JI Shichao,QU Xinghe,SONG Qingbin,et al. A lightweight insulator defect detection algorithm based on the improved YOLOv5[J].Zhejiang Electric Power,2023,42(12):64-72.
• [4]杨辉金.基于图像处理的接触网绝缘子裂纹和定位支座检测[D].成都：西南交通大学，2017.YANG Huijin.Detection of cracks and locating bearings of OCS insulators based on image processing[D].Chengdu:Southwest Jiaotong University,2017.
• [5]郭威，赵晓鹏.输电线路中缺失绝缘子的检测与定位[J].太原科技大学学报，2021,42(2):116-122.GUO Wei,ZHAO Xiaopeng. Detection and location of missing insulators in transmission lines[J].Journal of Taiyuan University of Science and Technology,2021,42(2):116-122.
• [6] BENALI A A,AMROUCH M. Object detection using deep learning,CNNs and vision transformers:a review[J].IEEE Access,2023,11:35479-35516.
• [7] CHEN L C,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,2017,40(4):834-848.
• [8]刘思言，王博，高昆仑，等.基于R-FCN的航拍巡检图像目标检测方法[J].电力系统自动化，2019,43(13):162-168.LIU Siyan,WANG Bo,GAO Kunlun,et al.Object detection method for aerial inspection image based on regionbased fully convolutional network[J].Automation of Electric Power Systems,2019,43(13):162-168.
• [9]翟学明，李晓，翟羽佳.基于深度学习的架空输电导线缺陷检测方法研究[J].电网技术，2023,47(3):1022-1031.ZHAI Xueming,LI Xiao,ZHAI Yujia.Defect detection of overhead transmission wires based on deep learning[J].Power System Technology,2023,47(3):1022-1031.
• [10]张烨，高玉菡，黄新波，等.基于Fine Mask RCNN的110~220 kV输电铁塔涉鸟故障识别与评估[J].电网技术，2022,46(6):2132-2144.ZHANG Ye,GAO Yuhan,HUANG Xinbo,et al. Fault identification and evaluation of tower wading in 110~220kV transmission line with fine mask RCNN[J]. Power System Technology,2022,46(6):2132-2144.
• [11]白晓静，谢雅祺，赵淼，等.基于局部特征深度信息的绝缘子小样本缺陷检测[J].电网技术，2024,48(2):740-752.BAI Xiaojing,XIE Yaqi,ZHAO Miao,et al.Few-shot insulator defect detection based on deep information of local features[J]. Power System Technology,2024,48(2):740-752.
• [12]应樱.基于改进Faster R-CNN的变电站设备缺陷检测算法研究[D].杭州：浙江大学，2021.YING Ying.Research on defect detection algorithm of substation equipment based on improved Faster R-CNN[D].Hangzhou:Zhejiang University,2021.
• [13]尹思宇.深度学习在变电站电力设备缺陷检测中的应用研究[D].成都：电子科技大学，2020.YIN Siyu.Research on application of deep learning in defect detection of substation power equipment[D].Chengdu:University of Electronic Science and Technology of China,2020.
• [14]徐海洋.变电站设备表面缺陷图像识别关键技术研究[D].南宁：广西大学，2021.XU Haiyang.Research on key technologies of image recognition for surface defects of substation equipment[D].Nanning:Guangxi University,2021.
• [15] LIU Z,LIN Y T,CAO Y,et al.Swin transformer:hierarchical vision transformer using shifted windows[C]//2021 IEEE/CVF International Conference on Computer Vision(ICCV),October 10-17,2021,Montreal,QC,Canada.IEEE,2021.
• [16] ZHOU H J,XIE X H,LAI J H,et al. Interactive twostream decoder for accurate and fast saliency detection[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR),June 13-19,2020,Seattle,WA,USA.IEEE,2020:9141-9150.
• [17] SONG Z,KANG X,WEI X,et al.Fsnet:focus scanning network for camouflaged object detection[J].IEEE Transactions on Image Processing,2023,32:2267-2278.
• [18] KINGMA D P,BA J,HAMMAD M M.Adam:a method for stochastic optimization[EB/OL].[2024-10-29].2014:1412.6980.https：//arxiv.org/abs/1412.6980v9.
• [19] RUSSAKOVSKY O,DENG J,SU H,et al. ImageNet large scale visual recognition challenge[J]. International Journal of Computer Vision,2015,115(3):211-252.
• [20] MARGOLIN R,ZELNIK-MANOR L,TAL A.How to evaluate foreground maps[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition. June 23-28,2014,Columbus,OH,USA.IEEE,2014:248-255.
• [21] FAN D P,CHENG M M,LIU Y,et al. Structuremeasure:a new way to evaluate foreground maps[C]//2017 IEEE International Conference on Computer Vision(ICCV),October 22-29,2017,Venice.IEEE,2017:4548-4557.
• [22] FAN D P,GONG C,CAO Y,et al.Enhanced-alignment measure for binary foreground map evaluation[EB/OL].[2024-10-29]2018:1805.10421. https：//arxiv. org/abs/1805.10421v2.
• [23] LIU J,YUAN M,HUANG X D,et al.DIPONet:Dualinformation progressive optimization network for salient object detection[J].Digital Signal Processing,2022,126:103425.
• [24] WEI J,WANG S H,HUANG Q M.F3Net:fusion,feedback and focus for salient object detection[J].Proceedings of the AAAI Conference on Artificial Intelligence,2020,34(7):12321-12328.
• [25] ZHUGE M,FAN D P,LIU N,et al.Salient object detection via integrity learning[J]. IEEE Trans Pattern Anal Mach Intell,2023,45(3):3738-3752.
• [26] YI Y G,ZHANG N Y,ZHOU W,et al.GPONet:a twostream gated progressive optimization network for salient object detection[J]. Pattern Recognition,2024,150:110330.