基于协同深度学习的二阶段绝缘子故障检测方法

doi:10.19595/j.cnki.1000-6753.tces.201320

摘要
图/表
参考文献
相关文章 (15)

全文: PDF (36065 KB) HTML
输出: BibTeX | EndNote (RIS)

摘要针对现有绝缘子故障检测模型受航拍图像中复杂背景干扰导致准确率低的问题,提出一种基于协同深度学习的二阶段绝缘子故障检测方法。该方法将全卷积网络（FCN）与YOLOv3目标检测算法相协同,第一阶段,利用FCN算法对航拍图像预处理,设计跳跃结构融合浅层图像特征与深层语义特征,构建8倍上采样的绝缘子分割模型,结合图像像素逻辑运算,实现绝缘子目标的初步分割,避免背景区域对绝缘子故障检测的干扰。在此基础上,第二阶段构建YOLOv3模型进行绝缘子故障检测,以深度神经网络Darknet-53作为特征提取器,借鉴特征金字塔思想,在三个尺度的输出张量上对绝缘子故障进行标记和类别预测,保证模型对不同尺寸的绝缘子故障准确检测。利用K-means++聚类算法优化YOLOv3的锚点框参数（Anchor Boxes）,进一步提升检测精度。实验结果表明,基于协同深度学习的二阶段方法能够有效克服复杂背景的干扰,在绝缘子故障检测中平均准确率（MAP）高达96.88%,较原始YOLOv3算法MAP值提升了4.65%。

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	王卓
	王玉静
	王庆岩
	康守强
	V.I.Mikulovich

关键词 ：绝缘子, 故障检测, 全卷积网络, YOLOv3, K-means++

Abstract：For the problem of low accuracy caused by complex background interference in aerial images of existing insulator fault detection models, a two-stage insulator fault detection method based on collaborative deep learning is proposed. The method combines fully convolutional networks (FCN) with YOLOv3 target detection algorithm. The first stage, the FCN algorithm is used to preprocess the aerial image. The jump structure is designed to fuse the shallow image features and deep semantic features, and an 8-fold up-sampling insulator segmentation model is constructed. Combining with the image pixel logic operation, the initial segmentation of insulator target can be realized to avoid the interference of background area on insulator fault detection. On this basis, the second stage, YOLOv3 model is constructed to detect the insulator faults. The deep neural network Darknet-53 is used as the feature extractor, and referring to the idea of feature pyramid, the insulator faults can be marked and classified on three scales of output tensors to ensure the accurate detection of insulator faults of different sizes. K-means++ clustering algorithm is used to optimize the anchor boxes parameters of YOLOv3 to further improve the detection accuracy. The experimental results show that the two-stage method based on collaborative deep learning can effectively overcome the interference of complex background, the mean average precision of insulator fault detection is as high as 96.88%, which is 4.65% higher than the original YOLOv3 algorithm.

Key words： Insulator fault detection fully convolutional networks YOLOv3 K-means++

收稿日期: 2020-09-27

PACS:

TM507

基金资助:黑龙江省自然科学基金（LH2019E058）、黑龙江省本科高校青年创新人才培养计划（UNPYSCT-2017091）和黑龙江省普通高校基本科研业务专项资金（LGYC2018JC022）资助项目

通讯作者: 王玉静女,1983年生,博士,副教授,研究方向为非平稳信号处理,故障诊断、状态评估与预测技术。E-mail：mirrorwyj@163.com

作者简介: 王卓男,1996年生,硕士研究生,研究方向为图像处理与目标识别、电力设备故障检测。E-mail：945861444@qq.com

引用本文:

王卓, 王玉静, 王庆岩, 康守强, V.I.Mikulovich. 基于协同深度学习的二阶段绝缘子故障检测方法[J]. 电工技术学报, 2021, 36(17): 3594-3604. Wang Zhuo, Wang Yujing, Wang Qingyan, Kang Shouqiang, V. I. Mikulovich. Two Stage Insulator Fault Detection Method Based on Collaborative Deep Learning. Transactions of China Electrotechnical Society, 2021, 36(17): 3594-3604.

链接本文:

https://dgjsxb.ces-transaction.com/CN/10.19595/j.cnki.1000-6753.tces.201320 https://dgjsxb.ces-transaction.com/CN/Y2021/V36/I17/3594

[1] 张倩, 王建平, 李帷韬. 基于反馈机制的卷积神经网络绝缘子状态检测方法[J]. 电工技术学报, 2019, 34(16): 3311-3321.
Zhang Qian, Wang Jianping, Li Weitao.Insulator state detection of convolutional neural networks based on feedback mechanism[J]. Transactions of China Electrotechnical Society, 2019, 34(16): 3311-3321.
[2] 马世伟, 张鑫, 范兴明, 等. 电力绝缘子检测方法及其应用现状[J]. 桂林电子科技大学学报, 2013, 33(6): 456-460.
Ma Shiwei, Zhang Xin, Fan Xingming, et al.Detection methods of electrical insulator and its application status[J]. Journal of Guilin University of Electronic Technology, 2013, 33(6): 456-460.
[3] 孙曙光, 李勤, 杜太行, 等. 基于一维卷积神经网络的低压万能式断路器附件故障诊断[J]. 电工技术学报, 2020, 35(12): 2562-2573.
Sun Shuguang, Li Qin, Du Taihang, et al.Fault diagnosis of accessories for the low voltage conventional circuit breaker based on one-dimensional convolutional neural network[J]. Transactions of China Electrotechnical Society, 2020, 35(12): 2562-2573.
[4] 李彩林, 张青华, 陈文贺, 等. 基于深度学习的绝缘子定向识别算法[J]. 电子与信息学报, 2020, 42(4): 1033-1040.
Li Cailin, Zhang Qinghua, Chen Wenhe, et al.Insulator orientation detection based on deep learning[J]. Journal of Electronics & Information Technology, 2020, 42(4): 1033-1040.
[5] Zhang Xinye, An Jubai, Chen Fangming.A method of insulator fault detection from airborne images[C]// IEEE 2010 Second WRI Global Congress on Intelligent Systems, Wuhan, 2010: 200-203.
[6] 朱邵成, 高清维, 卢一相, 等. 基于频率调谐的绝缘子识别与定位[J]. 电工技术学报, 2018, 33(23): 5573-5580.
Zhu Shaocheng, Gao Qingwei, Lu Yixiang, et al.Identification and location of insulator string based on frequency-tuned[J]. Transactions of China Electrote-chnical Society, 2018, 33(23): 5573-5580.
[7] 张桂南, 刘志刚. 基于角点匹配与谱聚类的接触网绝缘子破损/夹杂异物故障检测[J]. 仪器仪表学报, 2014, 35(6): 1370-1377.
Zhang Guinan, Liu Zhigang.Fault detection of catenary insulator damage / foreign material based on corner matching and spectral clustering[J]. Chinese Journal of Scientific Instrument, 2014, 35(6): 1370-1377.
[8] 金立军, 曹培, 胡娟. 可见光图像颜色特征与支持向量机相结合的绝缘子污秽状态识别方法[J]. 高压电器, 2015, 51(2): 1-7, 17.
Jin Lijun, Cao Pei, Hu Juan.Identification of insulator contamination grade combining color features of visual image with support vector machine[J]. High Voltage Apparatus, 2015, 51(2): 1-7, 17.
[9] 姜维, 张重生, 殷绪成. 基于深度学习的场景文字检测综述[J]. 电子学报, 2019, 47(5): 1152-1161.
Jiang Wei, Zhang Zhongsheng, Yin Xucheng.Deep learning based scene text detection: a survey[J]. Acta Electronica Sinica, 2019, 47(5): 1152-1161.
[10] 张慧, 王坤峰, 王飞跃. 深度学习在目标视觉检测中的应用进展与展望[J]. 自动化学报, 2017, 43(8): 1289-1305.
Zhang Hui, Wang Kunfeng, Wang Feiyue.Advances and perspectives on applications of deep learning in visual object detection[J]. Acta Automatica Sinica, 2019, 47(5): 1289-1305.
[11] 张顺, 龚怡宏, 王进军. 深度卷积神经网络的发展及其在计算机视觉领域的应用[J]. 计算机学报, 2019, 42(3): 453-482.
Zhang Shun, Gong Yihong, Wang Jinjun.The development of deep convolution neural network and its application on computer vision[J]. Chinese Journal of Computers, 2019, 42(3): 453-482.
[12] Girshick R, Donahue J, Darrell T, et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]//IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH, 2014: 580-587.
[13] Girshick R.Fast R-CNN[C]//IEEE International Conference on Computer Vision, Santiago, Chile, 2015: 1440-1448.
[14] 罗会兰, 陈鸿坤. 基于深度学习的目标检测研究综述[J]. 电子学报, 2020, 48(6): 1230-1239.
Luo Huilan, Chen Hongkun.Survey of object detection based on deep learning[J]. Acta Electronica Sinica, 2020, 48(6): 1230-1239.
[15] Ren Shaoqing, He Kaiming, Girshick R, et al.Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149.
[16] 林刚, 王波, 彭辉, 等. 基于改进Faster-RCNN的输电线巡检图像多目标检测及定位[J]. 电力自动化设备, 2019, 39(5): 213-218.
Lin Gang, Wang Bo, Peng Hui, et al.Multi-target detection and location of transmission line inspection image based on improved Faster-RCNN[J]. Electric Power Automation Equipment, 2019, 39(5): 213-218.
[17] Liu Wei, Anguelov D, Erhan D.SSD: single shot multi-box detector[C]//European Conference on Computer Vision, 2016: 21-37.
[18] Redmon J, Divvala S, Girshick R, et al.You only look once: unified, real-time object detec-tion[C]//IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, 2016: 779-788.
[19] 王旭红, 李浩, 樊绍胜, 等. 基于改进SSD的电力设备红外图像异常自动检测方法[J]. 电工技术学报, 2020, 35(增刊1): 302-310.
Wang Xuhong, Li Hao, Fan Shaosheng, et al.Infrared image anomaly automatic detection method for power equipment based on improved single shot multi box detection[J]. Transactions of China Electrotechnical Society, 2020, 35(S1): 302-310.
[20] Redmon J, Farhadi A. YOLOv3: an incremental improvement[J/OL]. [2018-04-08]. http://arXiuorg/ abs/1804.02767.
[21] 颜宏文, 陈金鑫. 基于改进YOLOv3的绝缘子串定位与状态识别方法[J]. 高电压技术, 2020, 46(2): 423-432.
Yan Hongwen, Chen Jinxin.Insulator string positioning and state recognition method based on improved YOLOv3 algorithm[J]. High Voltage Engineering, 2020, 46(2): 423-432.
[22] Lin Y, Goyal P, Girshick R, et al.Focal loss for dense object detection[C]//IEEE International Conference on Computer Vision, Venice, Italy, 2017: 2999-3007.