Detection of Common Foreign Matters on Power Grid Lines Based on Faster R-CNN

doi:10.16628/j.cnki.2095-8188.2021.01.010

Abstract

Abstract:

Foreign matters hanging on the power grid line and electrical equipment are easy to cause the contact short circuit between the lines which threatens the safety and stability of the power grid.In this paper,a high-precision object detection model based on Faster R-CNN was proposed for common foreign matters in power grid,such as kites,balloons,birds’ nests.Aiming at the problem of insufficient sample size of pictures,the data augmentation method is designed to effectively expand the data of the training set.Multi-scale processing of sample images is also carried out to improve the training speed of the model and the adaptability of the model to different scales.The accuracy and recall rate of the model can reach 93.12% and 94.75% respectively.It is proved that the target detection model based on Faster R-CNN trained by the training set with data enhancement has achieved better detection results.

Key words: foreign matters, convolutional neural network, data augmentation, object detection

CLC Number:

TM930

ZHU Hongzhi, SUN Zhen, LAN Qiaoqian, SHEN Liangyi, ZHANG Chong, HE Xing. Detection of Common Foreign Matters on Power Grid Lines Based on Faster R-CNN[J]. LOW VOLTAGE APPARATUS, 2021, 0(1): 58-63.

Figures/Tables 11

References 16

[1]	贺兴, 艾芊, 邱才明, 等. 泛在电力物联网数据挖掘体系建设综述及数据驱动认知框架探究[J]. 电器与能效管理技术, 2019(19):1-4.
[2]	张培铭. 智能低压电器技术研究[J]. 电器与能效管理技术, 2019(15):10-20.
[3]	GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]. CVPR,IEEE, 2014.
[4]	GIRSHICK R. Fast R-CNN[C]// 2015 IEEE International Conference on Computer Vision, 2015:1440-1448.
[5]	REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN:towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2015,39(6):1137-1149. doi: 10.1109/TPAMI.2016.2577031 pmid: 27295650
[6]	雷忠腾, 宋杰. 基于卷积神经网络模型Faster R-CNN的遥感影像目标识别研究[J]. 测绘与空间地理信息, 2021,44(1):149-151,155.
[7]	刘媛. 基于深度学习的遥感图像目标检测方法[J]. 科学技术创新, 2021(3):84-85.
[8]	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once:Unified,real-time object detection. Computer Vision & Pattern Recognition[C]. 2016.
[9]	LIU W, ANGUELOV D, ERHAN D, et al. SSD:single shot multiBox detector:European Conference on Computer Vision. Springer International Publishing[C]. 2016.
[10]	汤踊, 韩军, 魏文力, 等. 深度学习在输电线路中部件识别与缺陷检测的研究[J]. 电子测量技术, 2018,41(6):60-65.
[11]	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, 39(5):213-218.
[12]	虢韬, 杨恒, 时磊, 等. 基于Faster RCNN的绝缘子自爆缺陷识别[J]. 电瓷避雷器, 2019(3):183-189.
[13]	SHIN H C, ROTH H R, GAO M, et al. Deep convolutional neural networks for computer-aided detection:CNN architectures,dataset characteristics and transfer learning[J]. IEEE Transactions on Medical Imaging, 2016,35(5):1285-1298. doi: 10.1109/TMI.2016.2528162 pmid: 26886976
[14]	盛恒, 黄铭, 杨晶晶. 基于Faster R-CNN和IoU优化的实验室人数统计与管理系统[J]. 计算机应用, 2019,39(6):1669-1674.
[15]	BALLESTER P L, ARAUJO R M. On the performance of GoogLeNet and AlexNet applied to sketches[C]// AAAI.AAAI Press, 2016.
[16]	张烨, 许艇, 冯定忠, 等. 基于难分样本挖掘的快速区域卷积神经网络目标检测研究[J]. 电子与信息学报, 2019,41(6):1496-1502.

旋转角度/(°)	亮度	色彩饱和度
30°	原图像×0.5	原图像×0.5
150°	原图像×1.5	原图像×1.5

目录	TP	FP	FN	准确率/%	召回率/%
扩充前	102	22	18	82.26	85.00
扩充后	379	28	21	93.12	94.75

算法	准确率/%	召回率/%	速度/(ms·张^-1)
基于Faster R-CNN	93.12	94.75	145
基于YOLOv3	82.46	85.74	93