| 646 | 0 | 98 |
| 下载次数 | 被引频次 | 阅读次数 |
针对绝缘子多类型缺陷检测速度慢、检测精度低的问题，提出一种用于输电线路绝缘子多缺陷检测的轻量级网络(Multi-Defect Detection Network, MDDNet),该算法主要针对绝缘子电弧烧伤和绝缘子伞裙破损的多类型绝缘子缺陷联合检测。基于Ghost-C2f模块构建GC-Darknet53特征提取网络，增强特征提取能力并较少特征冗余；引入三尺度融合(Tri-Fusion)机制构建新型TF-Neck颈部网络，充分融合深层语义信息与浅层的细粒度信息，提高小目标缺陷检测精度；选用结构相似性交并比(Structural Similarity Intersection over Union, SIoU)损失函数提高模型定位能力。实验结果表明，提出的MDDNet模型平均精度均值(mean Average Precision mAP)达到92.1%,与YOLOv5相比，在参数量减少了20%的情况下mAP提升了3.0%,与其他现有一阶段算法相比，MDDNet算法检测速度达到86.1帧/秒，能够在保证轻量化的同时提高检测精度，满足绝缘子多缺陷检测的应用需求。
Abstract：To solve the problems of slow detection speed and low detection accuracy of multi-type defects of insulators, a Multi-Defect Detection Network(MDDNet) lightweight network for multi-defect detection of transmission line insulators is proposed. This algorithm mainly targets at the arc burns and shed breakage of insulators for joint detection of multi-type insulator defects. Firstly, the GC-Darknet53 feature extraction network is constructed based on the Ghost-C2f module to enhance the ability of feature extraction and reduce feature redundancy. Then, a Three-scale Fusion(Tri-Fusion) mechanism is introduced to construct a new TF-Neck neck network, which fully integrates the deep semantic information with the shallow fine-grained information to improve the accuracy of defect detection for small targets. Finally, the Structural Similarity Intersection over Union(SIoU) loss function is chosen to improve localization ability of the model. Experimental results show that the mean Average Precision(mAP) of the proposed MDDNet model reaches 92.1%. Compared with YOLOv5, the mAP improves by 3.0% with a 20% reduction in the amount of parameters. Compared with other existing one-stage algorithms, the MDDNet algorithm achieves a detection speed of 86.1 frame/s, which can improve detection precision and ensure lightweight, satisfying the application requirements of insulator multi-defect detection.
[1] YANG K,ZHANG Y,ZHANG X,et al.YOLOX with CBAM for Insulator Detection in Transmission Lines[J].Multimedia Tools and Applications,2023,83:43419-43437.
[2] LIU C,YI W T,LIU M,et al.A Lightweight Network Based on Improved YOLOv5s for Insulator Defect Detection[J].Electronics,2023,12(20):4292.
[3] SAMPEDRO C,RODRIGUEZ-VAZQUEZ J,RODRIGUEZ-RAMOS A,et al.Deep Learning-based System for Automatic Recognition and Diagnosis of Electrical Insulator Strings[J].IEEE Access,2019,7:101283-101308.
[4] 彭向阳，刘洋，王柯，等.利用卷积神经网络进行绝缘子自动定位[J].武汉大学学报(信息科学版),2019,44(4):563-569.
[5] 苟军年，杜愫愫，刘力.基于改进掩膜区域卷积神经网络的输电线路绝缘子自爆检测[J].电工技术学报，2023,38(1):47-59.
[6] TAN P,LI X F,DING J,et al.Mask R-CNN and Multifeature Clustering Model for Catenary Insulator Recognition and Defect Detection[J].Journal of Zhejiang University-SCIENCE A,2022,23(9):745-756.
[7] 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.
[8] NIU S S,ZHOU X S,ZHOU D S,et al.Fault Detection in Power Distribution Networks Based on Comprehensive-YOLOv5[J].Sensors,2023,23(14):6410.
[9] WANG Y X,YAN J,SUN Q F,et al.ShuffleNet-based Comprehensive Diagnosis for Insulation and Mechanical Faults of Power Equipment[J].High Voltage,2021,6(5):861-872.
[10] 王佰川，王聪.基于改进YOLOv4的配电线路绝缘子与避雷器快速检测研究[J].电瓷避雷器，2023(3):166-174.
[11] ZHANG S H,QU C N,RU C Y,et al.Multi-objects Re-cognition and Self-explosion Defect Detection Method for Insulators Based on Lightweight GhostNet-YOLOV4 Model Deployed Onboard UAV[J].IEEE Access,2023,11:39713-39725.
[12] 李斌，曾筠婷，朱新山，等.基于多尺度上下文感知的绝缘子缺陷检测网络[J].高电压技术，2022,48(8):2905-2914.
[13] 贾晓芬，于业齐，郭永存，等.航拍绝缘子自爆缺陷的轻量化检测方法[J].高电压技术，2023,49(1):294-300.
[14] 王道累，张世恒，袁斌霞，等.基于改进YOLOv5的轻量化玻璃绝缘子自爆缺陷检测研究[J].高电压技术，2023,49(10):4382-4390.
[15] QIU Z B,ZHU X,LIAO C B,et al.Detection of Transmission Line Insulator Defects Based on an Improved Lightweight YOLOv4 Model[J].Applied Science,2022,12(3):1207.
[16] GEVORGYAN Z.SIoU Loss:More Powerful Learning for Bounding Box Regression [EB/OL].(2022-05-25)[2023-10-15].https://arxiv.org/abs/2205.12740.
[17] HOWARD A,SANDLER M,CHEN B,et al.Searching for MobileNetV3[C]//Proceedings of 2019 IEEE/CVF International Conference on Computer Vision (ICCV).Seoul:IEEE,2019:1314-1324.
[18] HAN K,WANG Y H,TIAN Q,et al.GhostNet:More Features from Cheap Operations[C]//Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Seattle:IEEE,2020:1577-1586.
[19] REIS D,KUPEC J,HONG J,et al.Real-time Flying Object Detection with YOLOv8 [EB/OL].(2023-05-17)[2023-10-15].https://arxiv.org/abs/2305.09972v1.
[20] CHEN J R,KAO S H,HE H,et al.Run,Don't Walk:Chasing Higher FLOPS for Faster Neural Networks[C]//Proceedings of 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Vancouver:IEEE,2023:12021-12031.
[21] 陈彬，刘华洲，李勃铖，等.基于YOLOv5算法的无人机巡检图像中绝缘子单目测距方法[J/OL].无线电工程，1-9[2023-11-01].https://kns.cnki.net/kcms2/article/abstract?v=JgtjNxUAsgeSgfx7sjjev0kp-iwmyNS1 jTb3hn7rVzHdiOeTwtHEG0WE_kzxnVoiQjlkS5r4PiI_EfR CK3iHB33UNITjTtJ4LkRRvtxiii6XYEf8iQIaH2vQSe2b5C ZwaXZgFtFhvUUhycTY22xhZdULiq6dk8DH&uniplatform= NZKPT&language=CHS&version=LYDG.
[22] TSUNG-YI L,PRIYA G,ROSS G,et al.Focal Loss for Dense Object Detection[C]//Proceedings of 2017 IEEE International Conference on Computer Vision (ICCV).Venice:IEEE,2017:2999-3007.
[23] ZHENG G,SONGTAO L,FENG W,et al.YOLOX:Exceeding YOLO Series in 2021 [EB/OL].(2021-08-06) [2023-10-15].https://arxiv.org/abs/2107.08430.
[24] WANG C Y,BOCHKOVSKIY A,LIAO H Y M.YOLOv7:Trainable Bag-of-freebies Sets New State-of-the-art for Real-time Object Detectors[C]//Proceedings of 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Vancouver:IEEE,2023:7464-7475.
[25] 唐靓，余明慧，武明虎，等.基于改进YOLOv5的绝缘子缺陷检测算法[J].华中师范大学学报(自然科学版),2022,56(5):771-780.
[26] 伍箴燎，吴正平，孙水发.基于改进YOLOv5算法的绝缘子多缺陷检测[J/OL].高压电器，1-11[2023-10-15].http://kns.cnki.net/kcms/detail/61.1127.TM.20230814.1815.002.html.
基本信息:
中图分类号:TP391.41;TM216
引用信息:
[1]文斌,胡一鸣,彭顺,等.一种用于输电线路绝缘子多缺陷检测的轻量级网络[J].无线电工程,2024,54(10):2469-2477.
基金信息:
国家自然科学基金(62273200,61876097); 湖北省输电线路工程技术研究中心(三峡大学)开放研究项目(2022KXL03)~~