基于Faster RCNN的绝缘子自爆缺陷识别
|
摘要
绝缘子是保障输电线正常运行的重要部件,而传统通过人工目视判断的绝缘子缺陷检测方式耗材耗力,近年来得到大力推广的无人机电力巡检方式在图像中快速定位并且找出缺陷绝缘子自动化程度不高,效率较低。针对无人机电力巡检方式提出基于深度学习的Faster RCNN方法,识别无人机图像中的绝缘子,并且对其进行缺陷判别。首先,进行绝缘子的样本采集,采集的样本保证种类丰富、数量足够。其次,利用采集的样本训练Faster-RCNN网络模型,对待检测图像进行识别,确定绝缘子所在具体位置。最后,对提取出绝缘子进行图像分割,通过绝缘子片之间的距离判别该绝缘子串是否存在自爆缺陷。以实际无人机航拍图像为实验数据验证本文绝缘子自爆缺陷识别,结果表明,该方法在复杂背景下能够精确有效地识别出绝缘子串,并且准确判断出该绝缘子是否存在自爆缺陷,具有很强的鲁棒性和实用性。
Insulator is an important component to ensure the normal operation of electric power wire.The traditional method of artificial visualization is not only time-consuming,but also consumes financial resources.The UAV power inspection mode has been vigorously promote in recent years,but the degree of automation is not high and low efficiency in the image of the rapid positioning and identify.A new method based on the deep learning is proposed to identify the insulator and the defect type in UAV images.First of all,the extensive and sufficient samples of the insulators are collected.Secondly,the Faster RCNN network model is trained by the collected samples,and the characteristics of the insulator are studied.The network could identify the detection images to determine the exact location of the insulator.Finally,the whole insulator string is extracted,and the insulator piece of the insulators is extracted.The distance between the insulators is used to determine whether the insulator string has a self-explosion.The experimental results show that the proposed method can accurately and effectively identify the insulators in the complex background,and can accurately determine whether it is damaged.
Insulator is an important component to ensure the normal operation of electric power wire.The traditional method of artificial visualization is not only time-consuming,but also consumes financial resources.The UAV power inspection mode has been vigorously promote in recent years,but the degree of automation is not high and low efficiency in the image of the rapid positioning and identify.A new method based on the deep learning is proposed to identify the insulator and the defect type in UAV images.First of all,the extensive and sufficient samples of the insulators are collected.Secondly,the Faster RCNN network model is trained by the collected samples,and the characteristics of the insulator are studied.The network could identify the detection images to determine the exact location of the insulator.Finally,the whole insulator string is extracted,and the insulator piece of the insulators is extracted.The distance between the insulators is used to determine whether the insulator string has a self-explosion.The experimental results show that the proposed method can accurately and effectively identify the insulators in the complex background,and can accurately determine whether it is damaged.
引文
[1]林萍,王黎,罗林.基于SIFT的高速列车车顶瓷瓶识别方法[J].机车电传动,2012(5):87-89,92.LIN Ping,WANG Li,LUO Lin.SIFT-based porcelain vessel recognition method for high speed trains[J].Electric Drive for Locomotive,2012(5):87-89,92.
[2]王士成.基于学习算法受电弓瓷瓶定位检测方法的研究[D].成都:西南交通大学,2011.
[3]李和明,王胜辉,律方成,等.基于紫外成像的污秽绝缘子放电特性研究[J].华北电力大学学报:自然科学版,2010,37(3):1-6.LI Heming,WANG Shenghui,LV Fangcheng,et al.Study on discharge characteristics of contaminated insulator based on ultraviolet imaging[J].Journal of North China Electric Power University,2010,37(3):1-6.
[4]姚春羽,金立军,闫书佳.电网巡检图像中绝缘子的识别[J].系统仿真学报,2012,24(9):1818-1822.YAO Chunyu,JIN Lijun,YAN Shujia.Recognition of insulator string in power grid patrol images[J].Journal of System Simulation,2012,24(9):1818-1822.
[5]Zhang X,An J,Chen F.A Method of Insulator Fault Detection from Airborne Images[C]//Wri Global Congress on Intelligent Systems.IEEE Computer Society,2010:200-203.
[6]林聚财,韩军,陈舫明,等.基于彩色图像的玻璃绝缘子缺陷诊断[J].电网技术,2011(1):127-133.LIN Jucai,HAN Jun,CHEN Fangming,et al.Diagnosis of glass insulator defects based on color images[J].Power System Technology,2011(1):127-133.
[7]肖贤,周步祥,林楠,等.结合H-S变换和BSVM的高压输电线路故障识别[J].电力系统及其自动化学报,2015(2):70-76.XIAO Xian,ZHOU Buxiang,LIN Nan,et al.Fault identification of high-voltage transmission lines combining H-S transform and BSVM[J].Proceedings of the CSU-EPSA,2015(2):70-76.
[8]蔡双.基于Elman神经网络的输电线路故障识别[D].南昌:华东交通大学,2014.
[9]束洪春,安娜,董俊,等.高压直流输电线路故障识别的分形算法[J].电力系统自动化,2012,36(12):49-54,106.SHU Hongchun,AN Na,DONG Jun,et al.Fractal algorithm of faults identification for HVDC transmission lines[J].Automation of Electric Power Systems,2012,36(12):49-54,106.
[10]林圣,何正友,钱清泉.基于人工免疫算法的输电线路故障类型识别新方法[J].电力系统保护与控制,2011,39(11):95-100,137.LIN Sheng,HE Zhengyou,QIAN Qingquan.A novel approach of fault type recognition of transmission lines based on artificial immune algorithm[J].Power System Protection and Control,2011,39(11):95-100,137.
[11]李东敏,刘志刚,蔡军,等.基于多小波包系数熵和人工神经网络的输电线路故障类型识别方法[J].电网技术,2008(24):65-69.LI Dongmin,LIU Zhigang,CAI Jun,et al.Fault type identification method of transmission line based on multiwavelet packet coefficient entropy and artificial neural network[J].Power System Technology,2008(24):65-69.
[12]刘秀军,焦健,周会峰,等.基于故障行波传输有向树的网络定位算法[J].电力科学与技术学报,2016(4):58-64.LIU Xiujun,JIAO Jian,ZHOU Huifeng,et al.Network location algorithm based on fault traveling wave transmission directed tree[J].Journal of Electric Power Science and Technology,2016(4):58-64.
[13]孙雅明,王俊丰.基于分形理论的输电线路故障类型识别新方法[J].电力系统自动化,2005(12):23-28.SUN Yaming,WANG Junfeng.A new method for fault type identification of transmission lines based on fractal theory[J].Automation of Electric Power Systems,2005(12):23-28.
[14]GIRSHICK R,DONAHUE J,DARRELL T,et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]2014 IEEE CONFERENCE ONCOMPUTER VISION AND PATTERN RECOGNITION(CVPR).[S.l.]:[s.n.],2014:580-587.
[15]Girshick R.Fast r-cnn[C]//Proceedings of the IEEEInternational Conference on Computer Vision.2015:1440-1448.
[16]何东,刘瑞叶.基于主成分分析的神经网络动态集成风功率超短期预测[J].电力系统保护与控制,2013,41(4):50-54.HE Dong,LIU Ruishe.Ultra-short-term wind power prediction using ANN ensemble based on the principal components analysis[J].Power System Protection and Control,2013,41(4):50-54.
[2]王士成.基于学习算法受电弓瓷瓶定位检测方法的研究[D].成都:西南交通大学,2011.
[3]李和明,王胜辉,律方成,等.基于紫外成像的污秽绝缘子放电特性研究[J].华北电力大学学报:自然科学版,2010,37(3):1-6.LI Heming,WANG Shenghui,LV Fangcheng,et al.Study on discharge characteristics of contaminated insulator based on ultraviolet imaging[J].Journal of North China Electric Power University,2010,37(3):1-6.
[4]姚春羽,金立军,闫书佳.电网巡检图像中绝缘子的识别[J].系统仿真学报,2012,24(9):1818-1822.YAO Chunyu,JIN Lijun,YAN Shujia.Recognition of insulator string in power grid patrol images[J].Journal of System Simulation,2012,24(9):1818-1822.
[5]Zhang X,An J,Chen F.A Method of Insulator Fault Detection from Airborne Images[C]//Wri Global Congress on Intelligent Systems.IEEE Computer Society,2010:200-203.
[6]林聚财,韩军,陈舫明,等.基于彩色图像的玻璃绝缘子缺陷诊断[J].电网技术,2011(1):127-133.LIN Jucai,HAN Jun,CHEN Fangming,et al.Diagnosis of glass insulator defects based on color images[J].Power System Technology,2011(1):127-133.
[7]肖贤,周步祥,林楠,等.结合H-S变换和BSVM的高压输电线路故障识别[J].电力系统及其自动化学报,2015(2):70-76.XIAO Xian,ZHOU Buxiang,LIN Nan,et al.Fault identification of high-voltage transmission lines combining H-S transform and BSVM[J].Proceedings of the CSU-EPSA,2015(2):70-76.
[8]蔡双.基于Elman神经网络的输电线路故障识别[D].南昌:华东交通大学,2014.
[9]束洪春,安娜,董俊,等.高压直流输电线路故障识别的分形算法[J].电力系统自动化,2012,36(12):49-54,106.SHU Hongchun,AN Na,DONG Jun,et al.Fractal algorithm of faults identification for HVDC transmission lines[J].Automation of Electric Power Systems,2012,36(12):49-54,106.
[10]林圣,何正友,钱清泉.基于人工免疫算法的输电线路故障类型识别新方法[J].电力系统保护与控制,2011,39(11):95-100,137.LIN Sheng,HE Zhengyou,QIAN Qingquan.A novel approach of fault type recognition of transmission lines based on artificial immune algorithm[J].Power System Protection and Control,2011,39(11):95-100,137.
[11]李东敏,刘志刚,蔡军,等.基于多小波包系数熵和人工神经网络的输电线路故障类型识别方法[J].电网技术,2008(24):65-69.LI Dongmin,LIU Zhigang,CAI Jun,et al.Fault type identification method of transmission line based on multiwavelet packet coefficient entropy and artificial neural network[J].Power System Technology,2008(24):65-69.
[12]刘秀军,焦健,周会峰,等.基于故障行波传输有向树的网络定位算法[J].电力科学与技术学报,2016(4):58-64.LIU Xiujun,JIAO Jian,ZHOU Huifeng,et al.Network location algorithm based on fault traveling wave transmission directed tree[J].Journal of Electric Power Science and Technology,2016(4):58-64.
[13]孙雅明,王俊丰.基于分形理论的输电线路故障类型识别新方法[J].电力系统自动化,2005(12):23-28.SUN Yaming,WANG Junfeng.A new method for fault type identification of transmission lines based on fractal theory[J].Automation of Electric Power Systems,2005(12):23-28.
[14]GIRSHICK R,DONAHUE J,DARRELL T,et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]2014 IEEE CONFERENCE ONCOMPUTER VISION AND PATTERN RECOGNITION(CVPR).[S.l.]:[s.n.],2014:580-587.
[15]Girshick R.Fast r-cnn[C]//Proceedings of the IEEEInternational Conference on Computer Vision.2015:1440-1448.
[16]何东,刘瑞叶.基于主成分分析的神经网络动态集成风功率超短期预测[J].电力系统保护与控制,2013,41(4):50-54.HE Dong,LIU Ruishe.Ultra-short-term wind power prediction using ANN ensemble based on the principal components analysis[J].Power System Protection and Control,2013,41(4):50-54.