面向电力系统的多粒度隐患检测方法

徐晓华; 钱平; 王一达; 周昕悦; 徐汉麟; 徐李冰

doi:10.13700/j.bh.1001-5965.2020.0491

面向电力系统的多粒度隐患检测方法

doi: 10.13700/j.bh.1001-5965.2020.0491

1.
国网浙江省电力有限公司杭州供电公司, 杭州 310020
2.
国网浙江省电力有限公司, 杭州 310007

基金项目:

国网浙江省电力有限公司科技项目 5211HZ19014U

详细信息

作者简介:
徐晓华  男, 硕士, 高级工程师。主要研究方向: 电力信息及通信管理

钱平  男, 硕士, 高级工程师。主要研究方向: 电力系统自动化

王一达  男, 硕士, 高级工程师。主要研究方向: 电力系统自动化

周昕悦  女, 硕士, 工程师。主要研究方向: 电力信息系统

徐汉麟  男, 工程师。主要研究方向: 电力信息系统

徐李冰  男, 工程师。主要研究方向: 网络信息技术

通讯作者:
王一达, E-mail: wang_yida@zj.sgcc.com.cn

中图分类号: TP391
计量
- 文章访问数: 492
- HTML全文浏览量: 168
- PDF下载量: 53
- 被引次数: 0
出版历程
- 收稿日期: 2020-09-02
- 录用日期: 2020-09-04
- 网络出版日期: 2021-03-20

Multi-granularity hazard detection method for electrical power system

1.
State Grid Zhejiang Electric Power Company Hangzhou Power Supply Company, Hangzhou 310020, China
2.
State Grid Zhejiang Electric Power Company, Hangzhou 310007, China

Funds:

Science and Technology Project of State Grid Zhejiang Electric Power Company 5211HZ19014U

More Information

Corresponding author: WANG Yida, E-mail: wang_yida@zj.sgcc.com.cn

摘要

摘要:
由于电力系统的安全问题往往会造成严重的经济或社会影响，隐患检测已成为电力系统不可或缺的重要环节。随着人工智能领域的发展，基于深度学习的智能化电力系统隐患检测技术逐渐得到越来越多的关注。但目前的方法大多只是单一地考虑图像的全局特征或局部特征，无法全面彻底表征图像，进而难以捕捉电力领域尤其室外复杂背景下的隐患检测。为此，基于深度学习技术，提出了一种面向电力系统的多粒度隐患检测方法MGNet。通过引入图像的多粒度信息，构建全局和局部网络，进行多粒度级检测；并通过不同粒度级检测结果的协作式融合，增强检测的全面性。在杆塔连接金具隐患和线路通道机械隐患2个数据集上进行了实验比较和分析，对所提模型的检测性能进行评估。通过与现有最优隐患检测基准方法相比，所提方法在2种不同数据集上的平均精度均值分别提升了2.74%和2.77%，验证了模型的有效性。
- 隐患检测 /
- 多粒度信息 /
- 协作式融合 /
- 深度学习 /
- 电力系统
Abstract:
As the security hazard of the electrical power system can lead to serious economic damage and social impacts, the potential hazard detection has become an indispensable part for electrical power system. With the advances of artificial intelligence, intelligent deep learning based hazard detection methods for electrical power system have emerged. Although the existing methods have made promising progress, most of them only consider the global or local features of the image, which cannot thoroughly characterize the imageand accurately conduct the hazard detection in the context of the electrical power system especially for the complex outdoor background. In the light of this, in this paper, we present a multi-granularity hazard detection network MGNet for the electrical power system. To be specific, we explore the multi-granularity representation of images with both the global and local representation learning networks. Based on that, we conduct the hazard detection at different granularity levels and finally collaboratively fuse the detection results to fulfill the precise hazard detection. Extensive experiments on two real-world datasets of hazard(i.e., tower connection fitting hazard dataset and transmission line channel mechanical hazard dataset) demonstrate the superiority of the detection performance of the proposed model. In particular, the mean average precision is improved by 2.74% and 2.77% on two datasets, respectively, compared with the existing optimal hazard detection benchmark method.
- hazard detection /
- multi-granularity /
- collaborative fusion /
- deep learning /
- electrical power system

HTML全文

图 1 多粒度隐患检测网络结构

Figure 1. Structure of multi-granularity hazard detection network

下载: 全尺寸图片幻灯片

图 2 区域建议网络结构

Figure 2. Structure of region proposal network

下载: 全尺寸图片幻灯片

图 3 代价函数收敛图

Figure 3. Cost function convergence graph

下载: 全尺寸图片幻灯片

图 4 MGNet和YOLOv4在杆塔连接金具隐患数据集中各隐患类别表现

Figure 4. Performance of MGNet and YOLOv4 on different categories of tower connection fitting hazard datasets

下载: 全尺寸图片幻灯片

图 5 MGNet和YOLOv4在线路通道机械隐患数据集中各隐患类别表现

Figure 5. Performance of MGNet and YOLOv4 on different categories of transmission line channel mechanical hazard datasets

下载: 全尺寸图片幻灯片

图 6 MGNet及其消融衍生模型在线路通道机械隐患数据集中各隐患类别表现

Figure 6. Performance of MGNet and its ablation derived models on different categories of transmission line channel mechanical hazard datasets

下载: 全尺寸图片幻灯片

图 7 MGNet及其消融衍生模型在杆塔连接金具隐患数据集中各隐患类别表现

Figure 7. Performance of MGNet and its ablation derived models on different categories of tower connection fitting hazard datasets

下载: 全尺寸图片幻灯片

图 8 消融实验检测结果对照图(m=3)

Figure 8. Comparison of detection results from ablation experiment (m=3)

下载: 全尺寸图片幻灯片

图 9 图 8检测结果放大图

Figure 9. Magnification of detection results in Fig. 8

下载: 全尺寸图片幻灯片

图 10 切块尺度因子在两个数据集中对模型MGNet平均精度均值的影响

Figure 10. Effect of block scale factor on mean average precision of MGNet model in two datasets

下载: 全尺寸图片幻灯片

图 11 融合阈值在杆塔连接金具隐患数据集中对模型MGNet平均精度均值的影响

Figure 11. Effect of fusion threshold on mean average precision of MGNet model in tower connection fitting hazard datasets

下载: 全尺寸图片幻灯片

图 12 融合阈值在线路通道机械隐患数据集中对模型MGNet平均精度均值的影响

Figure 12. Effect of fusion threshold on mean average precision of MGNet model in transmission line channel mechanical hazard datasets

下载: 全尺寸图片幻灯片

表 1 符号表示

Table 1. Summary of main notations' representation

符号	含义
C	隐患标签数目
o_j	第j个隐患物体
c_j	o_j的隐患类别标签标识
t_j	o_j的位置表示
	图像全局特征图
	图像在第r个网格中局部特征图表示
	隐患物体的预测类别
	隐患物体的预测位置

下载: 导出CSV

表 2 杆塔连接金具隐患数据集规模

Table 2. Statistics of tower connection fitting hazard datasets

隐患类别	隐患物体数目
销钉缺损	6 232
销钉安装位置错误	3 272

下载: 导出CSV

表 3 线路通道机械隐患数据集规模

Table 3. Statistics of transmission line channel mechanical hazard datasets

隐患类别	隐患物体数目
卡车	7 073
推土机	7 579
起重机	15 796
挖掘机	18 568
泵车	4 808
水泥混合器	2 449
打桩机	1 336

下载: 导出CSV

表 4 不同隐患检测方法在2个数据集上的平均精度均值

Table 4. Mean average precision of different hazard detection methods on two datasets

模型	mAP/%
模型	杆塔连接金具隐患数据集	线路通道机械隐患数据集
SSD	19.99	60.93
Faster R-CNN	38.96	71.65
YOLOv3	40.59	61.32
Cascade R-CNN	41.21	68.50
YOLOv4	44.99	72.34
MGNet	47.73	75.11

下载: 导出CSV

表 5 多粒度隐患检测网络消融实验

Table 5. Ablation experiments of multi-granularity hazard detection network

模型	mAP/%
模型	杆塔连接金具隐患数据集	线路通道机械隐患数据集
MGNet	47.73	75.11
MGNet_NoGlobal	46.87	41.24
MGNet_NoLocal	38.96	71.65

下载: 导出CSV

参考文献(23)

[1]	ZHAI Y J, WANG D, ZHANG M L, et al. Fault detection of insulator based on saliency and adaptive morphology[J]. Multimedia Tools and Applications, 2017, 76(9): 12051-12064. doi: 10.1007/s11042-016-3981-2
[2]	PRASAD P S, RAO B P. LBP-HF features and machine learning applied for automated monitoring of insulators for overhead power distribution lines[C]//Proceedings of the International Conference on Wireless Communications, 2016: 808-812.
[3]	NORDENG I E, HASAN A, OLSEN D, et al. DEBC detection with deep learning[C]//Proceedings of the Scandinavian Conference. Berlin: Springer, 2017: 248-259.
[4]	王万国, 田兵, 刘越, 等. 基于RCNN的无人机巡检图像电力小部件识别研究[J]. 地球信息科学, 2017, 19(2): 256-263. https://www.cnki.com.cn/Article/CJFDTOTAL-DQXX201702014.htm WANG W G, TIAN B, LIU Y, et al. Study on the electrical devices detection in UAV images based on region based convolutional neural networks[J]. Journal of Geo-information Science, 2017, 19(2): 256-263(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-DQXX201702014.htm
[5]	柴玉梅, 员武莲, 王黎明, 等. 基于双注意力机制和迁移学习的跨领域推荐模型[J]. 计算机学报, 2020, 43(9): 1-20. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJX202010007.htm CHAI Y M, YUAN W L, WANG L M, et al. A cross-domain recommendation model based on dual attention mechanism and transfer learning[J]. Chinese Journal of Computers, 2020, 43(9): 1-20(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-JSJX202010007.htm
[6]	严星, 尤洪峰. Faster-RCNN电力输送塔检测算法[J]. 计算机仿真, 2020, 37(2): 135-139. doi: 10.3969/j.issn.1006-9348.2020.02.028 YAN X, YOU H F. Target detection algorithm for Faster-RCNN power transmission tower[J]. Computer Integrated Manufacturing Systems, 2020, 37(2): 135-139(in Chinese). doi: 10.3969/j.issn.1006-9348.2020.02.028
[7]	陈耀东, 李仁发, 李实英, 等. 面向目标检测与姿态估计的联合文法模型[J]. 计算机学报, 2014, 37(10): 2206-2217. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJX201410014.htm CHEN Y D, LI R F, LI S Y, et al. A combined grammar for object detection and pose estimation[J]. Chinese Journal of Computers, 2014, 37(10): 2206-2217(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-JSJX201410014.htm
[8]	张冬明, 靳国庆, 代锋, 等. 基于深度融合的显著性目标检测算法[J]. 计算机学报, 2019, 42(9): 2076-2086. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJX201909012.htm ZHANG D M, JIN G Q, DAI F, et al. Salient object detection based on deep fusion of hand-crafted features[J]. Chinese Journal of Computers, 2019, 42(9): 2076-2086(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-JSJX201909012.htm
[9]	SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[C]//Proceedings of the International Conference on Learning Representation, 2015.
[10]	GIRSHICK R B. Fast R-CNN[C]//Proceedings of the IEEE International Conference on Computer Vision. Piscataway: IEEE Press, 2015: 1440-1448.
[11]	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2017: 770-778.
[12]	陆永帅, 李元祥, 刘波, 等. 基于深度残差网络的高光谱遥感数据霾监测[J]. 光学学报, 2017, 37(11): 314-324. https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB201711037.htm LU Y S, LI Y X, LIU B, et al. Hyperspectral data haze monitoring based on deep residual network[J]. Acta Optica Sinica, 2017, 37(11): 314-324(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB201711037.htm
[13]	朱超平, 杨艺. 基于YOLO2和ResNet算法的监控视频中的人脸检测与识别[J]. 重庆理工大学学报(自然科学), 2018, 32(8): 170-175. https://www.cnki.com.cn/Article/CJFDTOTAL-CGGL201808027.htm ZHU C P, YANG Y. Face detection and recognition in monitoring video based on YOLO2 and ResNet algorithm[J]. Journal of Chongqing University of Technology (Natural Science), 2018, 32(8): 170-175(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-CGGL201808027.htm
[14]	徐龙壮, 彭力, 朱凤增. 多任务金字塔重叠匹配的行人重识别方法[J/OL]. 计算机工程, 2020(2020-02-11)[2020-08-01]. https://kns.cnki.net/kcms/detail/detail.aspx?dbcode=CAPJ&dbname=CAPJLAST&filename=JSJC20200209002&v=fpDLQvPDFGfsJ7%25mmd2BlY85YejXOe%25mmd2B%25mmd2FpNa24x8p%25mmd2FiBh3e5i3ALZRcuH3l1F0Q0lM4Rgp. XU L Z, PENG L, ZHU F Z. Person re-identification method based on multi-task pyramid overlapping matching[J/OL]. Computer Engineering, 2020(2020-02-11)[2020-08-01]. https://kns.cnki.net/kcms/detail/detail.aspx?dbcode=CAPJ&dbname=CAPJLAST&filename=JSJC20200209002&v=fpDLQvPDFGfsJ7%25mmd2BlY85YejXOe%25mmd2B%25mmd2FpNa24x8p%25mmd2FiBh3e5i3ALZRcuH3l1F0Q0lM4Rgp (in Chinese).
[15]	UIJLINGS J R R, VAN DE SANDE K E A, GEVERS T, et al. Selective search for object recognition[J]. International Journal of Computer Vision, 2013, 104(2): 154-171. doi: 10.1007/s11263-013-0620-5
[16]	李文璞, 谢可, 廖逍, 等. 基于Faster RCNN变电设备红外图像缺陷识别方法[J]. 南方电网技术, 2019, 13(12): 79-84. https://www.cnki.com.cn/Article/CJFDTOTAL-NFDW201912012.htm LI W P, XIE K, LIAO X, et al. Intelligent diagnosis method of infrared image for transformer equipment based on improved Faster RCNN[J]. China Southern Power Grid, 2019, 13(12): 79-84(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-NFDW201912012.htm
[17]	徐向前, 孙涛. Faster RCNN的交通场景下行人检测方法[J]. 软件导刊, 2020, 19(4): 67-70. https://www.cnki.com.cn/Article/CJFDTOTAL-RJDK202004013.htm XU X Q, SUN T. Pedestrian detection method in traffic scene based on Faster RCNN[J]. Software Guide, 2020, 19(4): 67-70(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-RJDK202004013.htm
[18]	李林升, 曾平平. 改进深度学习框架Faster-RCNN的苹果目标检测[J]. 机械设计与研究, 2019, 35(5): 24-27. https://www.cnki.com.cn/Article/CJFDTOTAL-JSYY201905011.htm LI L S, ZENG P P. Apple target detection based on improved Faster-RCNN framework of deep learning[J]. Machine Design and Research, 2019, 35(5): 24-27(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-JSYY201905011.htm
[19]	LIU W, ANGUELOV D, ERHAN D, et al. SSD: Single shot MultiBox detector[C]//European Conference on Computer Vision. Berlin: Springer, 2016: 21-37.
[20]	REN S Q, HE K M, GIRSHICK R B, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 39(6): 91-99. doi: 10.1109/TPAMI.2016.2577031
[21]	REDMON J, FARHADI A. YOLOv3: An incremental improvement[EB/OL]. (2018-04-08)[2020-08-01]. https://arxiv.org/abs/1804.02767.
[22]	CAI Z W, VASCONCELOS N. Cascade R-CNN: Delving into high quality object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2018: 6154-6162.
[23]	BOCHKOVSKIY A, WANG C Y, LIAO H Y M. YOLOv4: Optimal speed and accuracy of object detection[EB/OL]. (2020-04-23)[2020-08-01]. https://arxiv.org/abs/2004.10934.