基于改进FCOS算法的架空输电线路防振锤检测*

doi:10.11985/2023.01.016

电气工程学报 ›› 2023, Vol. 18 ›› Issue (1): 143-152.doi: 10.11985/2023.01.016

扫码分享

基于改进FCOS算法的架空输电线路防振锤检测^*

吴彤¹(), 李冰锋¹(), 费树岷²(), 连东辉³^,⁴()

1.河南理工大学电气工程与自动化学院焦作 454003
2.东南大学自动化学院南京 210096
3.郑州煤矿机械集团股份有限公司郑州 450016
4.郑煤机液压电控有限公司郑州 450016

收稿日期:2021-08-24 修回日期:2021-12-14 出版日期:2023-03-25 发布日期:2023-04-19
通讯作者: 吴彤，女，1997年生，硕士研究生。主要研究方向为目标检测。E-mail：2454905251@qq.com
作者简介:李冰锋，男，1979年生，博士，讲师。主要研究方向为目标检测。E-mail：libingfeng@hpu.edu.cn
费树岷，男，1961年生，博士，教授。主要研究方向为人工智能、非线性系统。E-mail：smfei@seu.edu.cn
连东辉，男，1986年生，中级工程师。主要研究方向为自动化。E-mail：liandonghui@dingtalk.com
基金资助:
*河南省煤矿智能开采技术创新中心支撑(2021YD01);河南理工大学博士基金(B2018-33);贵州省科技计划(黔科合重大专项字[2018]3003-1)

Anti-vibration Hammer Detection of Overhead Transmission Lines Based on Improved FCOS Algorithm

WU Tong¹(), LI Bingfeng¹(), FEI Shumin²(), LIAN Donghui³^,⁴()

1. School of Electrical Engineering and Automation, Henan Polytechnic University, Jiaozuo 454003
2. School of Automation, Southeast University, Nanjing 210096
3. Zhengzhou Coal Mine Machinery Group Co., Ltd., Zhengzhou 450016
4. Zheng Coal Machine Hydraulic Electric Control Co., Ltd., Zhengzhou 450016

Received:2021-08-24 Revised:2021-12-14 Online:2023-03-25 Published:2023-04-19

摘要/Abstract

摘要：

防振锤是架空输电线路系统中一种重要的电气设备，对防止架空线路因风吹而发生周期性疲劳破坏具有重要意义。航拍图像中，防振锤具有尺寸较小、形态各异、背景复杂多变、检测难度较大等问题。针对这些问题，采用单阶段全卷积目标检测网络(Fully convolutional one-stage object detection，FCOS)来进行架空输电线路防振锤检测。为了提高检测精度，将FCOS特征提取层的各个特征点看作随机变量，用各阶中心矩的组合表达其随机分布，并在此基础上提出了一种基于各阶中心矩的空间注意力机制，来准确描述图像特征的权重分布。试验结果表明，改进后的FCOS在不同阈值下的平均检测精度均高于原始的FCOS，当阈值为0.5时，平均检测精度达到94.9%。同时，该方法在不同阈值下的平均检测精度，大大超过了其他主流的注意力机制。

关键词: 架空输电线路, FCOS, 防振锤检测, 矩特征, 空间注意力机制

Abstract:

Anti-vibration hammer is an important electrical equipment in overhead transmission line system. It is of great significance to prevent the periodic fatigue failure of overhead line due to wind blowing. In aerial images, the vibration hammer is small in size, different in shape, complex and changeable in background, and difficult to detect. To solve this problem, fully one-stage convolution target detection network(FCOS) is used to detect the vibration hammer of overhead transmission lines. In order to improve the detection accuracy, each feature point of FCOS feature extraction layer is regarded as a random variable, and its random distribution is expressed by the combination of each order center moment. On this basis, a spatial attention mechanism based on the central moments of each order is proposed to accurately describe the weight distribution of image features. The experimental results show that the average detection accuracy of the improved FCOS is higher than the original FCOS at different thresholds, when the threshold is 0.5, the average detection accuracy reaches 94.9%. At the same time, the average detection accuracy of this method under different threshold values is much better than other mainstream attention mechanisms.

Key words: Overhead transmission line, FCOS, anti-vibration hammer detection, moment feature, spatial attention mechanism

中图分类号:

TM726
TP183

吴彤, 李冰锋, 费树岷, 连东辉. 基于改进FCOS算法的架空输电线路防振锤检测^*[J]. 电气工程学报, 2023, 18(1): 143-152.

WU Tong, LI Bingfeng, FEI Shumin, LIAN Donghui. Anti-vibration Hammer Detection of Overhead Transmission Lines Based on Improved FCOS Algorithm[J]. Journal of Electrical Engineering, 2023, 18(1): 143-152.

图/表 14

图1

图2

图3

图4

表1

图5

图6

表2

表3

表4

图7

图8

图9

图10

参考文献 24

[1]	金立军, 闫书佳, 刘源. 基于类Haar特征与级联AdaBoost算法的防振锤识别[J]. 系统仿真学报, 2012, 24(9):1806-1809.
	JIN Lijun, YAN Shujia, LIU Yuan. Seismic hammer recognition based on Haar-like feature and cascading AdaBoost algorithm[J]. Journal of System Simulation, 2012, 24(9):1806-1809.
[2]	WU Haibin, XI Yanping, FANG Weimin, et al. Damper detection in helicopter inspection of power transmission line[C]// Fourth International Conference on Instrumentation and Measurement,Computer,Communication and Control,December 29,Harbin,China. IEEE, 2014:628-632.
[3]	朱铭武, 韩军, 陆冬明, 等. 自然场景中基于局部轮廓特征的对象识别方法[J]. 计算机工程与应用, 2016, 52(1):162-167.
	ZHU Mingwu, HAN Jun, LU Dongming, et al. Object recognition method based on local contour feature in natural scene[J]. Computer Engineering and Applications, 2016, 52(1):162-167.
[4]	GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]// Proceedings of 2014 IEEE Conference on Computer Vision and Pattern Recognition,Columbus,OH,USA. IEEE, 2014:580-587.
[5]	HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2014, 37(9):1904-1916.
[6]	GIRSHICK R. Fast R-CNN[C]// Proceedings of 2015 IEEE International Conference on Computer Vision,Santiago,Chile. IEEE, 2015:1440-1448.
[7]	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. doi: 10.1109/TPAMI.2016.2577031 pmid: 27295650
[8]	DAI Jifeng, LI Yi, HE Kaiming, et al. R-FCN:Object detection via region-based fully convolutional networks[C]// Advances in Neural Information Processing Systems. New York: Curran Associates Inc., 2016:379-387.
[9]	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once:Unified,real-time object detection[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, June 27-30,2016,Las Vegas,NV,USA. IEEE, 2016:779-788.
[10]	REDMON J, FARHADI A. YOLO9000:Better, faster, stronger[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), July 21-26,2017,Honolulu,HI,USA. IEEE, 2017:7263-7271.
[11]	REDMON J, FARHADI A. YOLOV3:An incremental improvement[J]. Computer Science, 2018, 4(1):1-6.
[12]	LIU W, ANGUELOV D, ERHAN D, et al. SSD:Single shot multibox detector[C]// European Conference on Computer Vision. Springer,Cham, 2016:21-37.
[13]	TIAN Zhi, SHEN Chunhua, CHEN Hao, et al. FCOS:Fully convolutional one-stage object detection[C]// Proceedings of the IEEE International Conference on Computer Vision,October 27-November 02,2019,Seoul,South Korea. IEEE, 2019:9626-9635.
[14]	BAO Wenxia, REN Yangxun, LIANG Dong, et al. Defect detection algorithm of anti-vibration hammer based on improved cascade R-CNN[C]// 2020 International Conference on Intelligent Computing and Human-Computer Interaction (ICHCI), December 04-06,2020,Sanya, China. IEEE, 2020:294-297.
[15]	LIU Xinyu, LIN Yating, JIANG Hao, et al. Slippag fault diagnosis of dampers for transmission lines based on faster R-CNN and distance constraint[J]. Electric Power Systems Research, 2021, 199:107449. doi: 10.1016/j.epsr.2021.107449
[16]	刘又维, 樊绍胜, 唐立军, 等. 基于关键点的输电线路防振锤滑移视觉检测[J]. 激光与光电子学进展. 2020, 57(20):279-285.
	LIU Youwei, FAN Shaosheng, TANG Lijun, et al. Visual detection of vibration damper slip in transmission lines based on key points[J]. Progress in Laser and Optoelectronics, 2020, 57(20):279-285.
[17]	HU J, SHEN L, ALBANIE S, et al. Squeeze-and- excitation networks[C]// IEEE Conference on Computer Vision and Pattern Recognition, 2018:7132-7141.
[18]	HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]// IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016:770-778.
[19]	GAO Zilin, XIE Jiangtao, WANG Qilong, et al. Global second-order pooling convolutional networks[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2019:3024-3033.
[20]	QIN Zequn, ZHANG Pengyi, WU Fei. FcaNet:Frequency channel attention networks[J]. Computer Vision and Pattern Recognition, 2021:763-772.
[21]	WANG Qilong, WU Banggu, ZHU Pengfei, et al. ECA-Net:Efficient channel attention for deep convolutional neural betworks[C]// IEEE Conference on Computer Vision and Pattern Recognition, 2020:11531-11539.
[22]	ZHANG Chi, LIN Guosheng, LIU Fayao, et al. CANet:Class-agnostic segmentation networks with iterative refinement and attentive few-shot learning[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2019:5217-5226.
[23]	WOO S, PARK J, LEE J, et al. CBAM:Convolutional block attention module[C]// The European Conference on Computer Vision, 2018:3-19.
[24]	LANDSKAPE D, NALAMADA T, ARASANIPAL A, et al. Rotate to attend:Convolutional triplet attention module[J]. Computer Vision and Pattern Recognition, 2021:3139-3148.

Layer1	Layer2	Layer3	Layer4	AP_0.5(%)	AP_0.6(%)	AP_0.7(%)	AP_0.8(%)	AP_average(%)
				94.2	89.4	75.4	34.3	73.33
				93.3	84.2	76.1	35.3	72.23
				94.2	90.3	76.9	37.8	74.80
				95.0	90.5	75.9	35.2	74.15
				94.4	89.8	77.5	33.6	73.83
				94.7	91.9	79.5	33.8	74.98
				93.1	88.7	73.7	34.9	72.60
				94.9	91.4	78.9	38.9	76.03

算法/阈值	参数	FLOPs	AP_0.5(%)	AP_0.6(%)	AP_0.7(%)	AP_0.8(%)	AP_average(%)
FCOS^[13]	23.508 0 M	156.58	88.0	77.2	52.5	11.6	57.33
FCOS+ECA-Net^[21]	23.508 1 M	156.89	87.2	73.1	53.1	28.5	60.48
FCOS+CA-Net^[22]	24.032 3 M	160.53	90.3	75.3	14.4	0.4	45.10
FCOS+CBAM^[23]	24.033 0 M	160.96	89.4	62.9	19.0	1.3	43.15
FCOS+Triplet Attention^[24]	23.508 3 M	157.43	94.4	89.3	72.2	20.9	69.20
FCOS+本文注意力机制	23.508 2 M	156.86	94.9	91.4	78.9	38.9	76.03

最大池化	全局平均池化	方差	偏度	AP_0.5(%)	AP_0.6(%)	AP_0.7(%)	AP_0.8(%)	AP_average(%)
				88.0	77.2	52.5	11.6	57.33
√	√			80.8	56.3	26.7	3.1	41.73
	√	√	√	92.0	86.9	63.8	14.9	64.40
√	√	√		93.2	88.6	68.7	26.3	69.20
√	√	√	√	94.9	91.4	78.9	38.9	76.03

基于改进FCOS算法的架空输电线路防振锤检测^*

Anti-vibration Hammer Detection of Overhead Transmission Lines Based on Improved FCOS Algorithm

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 14

参考文献 24

相关文章 0

编辑推荐

Metrics

本文评价