基于改进胶囊网络的接触网吊弦故障识别与定位

doi:10.19595/j.cnki.1000-6753.tces.191448

Abstract
Figure/Table
References
Related Citation (9)

Download: PDF (64822 KB) HTML (1 KB)
Export: BibTeX | EndNote (RIS)

Abstract The contact area of the catenary suspension is small and easy to be confused with the power line. The traditional fault identification algorithms have problems such as misidentification of the suspension, low recognition efficiency and inability to detect in real time. Compared with the traditional convolutional neural network (CNN), the capsule network (CapsNet) uses vectors as input, which can well retain the feature information such as the direction and angle of the target, and is more suitable for identifying the suspension under complex background. Therefore, a suspension fault recognition algorithm based on improved capsule network and CV model is proposed. The convolution layer of traditional 9×9 capsule network is simplified by 1×1 reduction layer and 3×3 convolution layer, and the optimization algorithm is adopted to shorten the training weight time. At the same time, the output retains the direction and angle, which can more accurately classify the chords of burns, current-carrying ring breaks, and falling off, and so on. The method can be applied to the suspension inspection image. In this way, the accuracy of the catenary suspension positioning is increased to 95%. Finally, compared with Back Propagation Neural Network and AlexNet designed by Alex Krizhevsky, the proposed method of suspension fault identification can identify the suspension from the complex background and find the suspension accurately and quickly. The efficiency of the transmission line intelligent inspection can be greatly improved.

Key words： Suspension fault identification improved capsule network

Received: 09 November 2019

PACS:

TM922.5

	Service

	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Bian Jianpeng
	Hao Jiaxing
	Zhao Shuai
	Li Fan
	Sun Xiaoyun

Cite this article:

Bian Jianpeng,Hao Jiaxing,Zhao Shuai等. Fault Identification and Location of Catenary Suspension Based on Improved Capsule Network[J]. Transactions of China Electrotechnical Society, 2020, 35(24): 5187-5196.

URL:

https://dgjsxb.ces-transaction.com/EN/10.19595/j.cnki.1000-6753.tces.191448 OR https://dgjsxb.ces-transaction.com/EN/Y2020/V35/I24/5187

[1] 刘满君, 程林, 黄道姗, 等. 基于运行可靠性理论的高可靠性供电路径搜索方法[J]. 电工技术学报, 2019, 34(14): 3004-3011.
Liu Manjun, Cheng Lin, Huang Daoshan, et al.A high reliability power supply path search method based on operation reliability theory[J]. Transactions of China Electrotechnical Society, 2019, 34(14): 3004-3011.
[2] 张血琴, 陈奎, 李瑞芳, 等. 高架桥段地铁接触网的改进防雷措施[J]. 高电压技术, 2016, 42(5): 1527-1534.
Zhang Xueqin, Chen Kui, Li Ruifang, et al.Improved lightning protection measures for metro contact network of viaduct section[J]. High Voltage Engin- eering, 2016, 42(5): 1527-1534.
[3] 李恩文, 王力农, 宋斌, 等. 基于改进模糊聚类算法的变压器油色谱分析[J]. 电工技术学报, 2018, 33(19): 4594-4602.
Li Enwen, Wang Linong, Song Bin, et al.Chromato- graphic analysis of transformer oil based on improved fuzzy clustering algorithm[J]. Transactions of China Electrotechnical Society, 2018, 33(19): 4594-4602.
[4] 程学珍, 朱晓林, 杜彦镔, 等. 基于神经模糊Petri网的高压断路器故障诊断研究[J]. 电工技术学报, 2018, 33(11): 2535-2544.
Cheng Xuezhen, Zhu Xiaolin, Du Yanbin, et al.A fault diagnosis of high voltage circuit breakers based on neural fuzzy Petri nets[J]. Transactions of China Electrotechnical Society, 2018, 33(11): 2535-2544.
[5] 姚海燕, 张静, 留毅, 等. 基于多尺度小波判据和时频特征关联的电缆早期故障检测和识别方法[J]. 电力系统保护与控制, 2015, 43(9): 115-123.
Yao Haiyan, Zhang Jing, Liu Yi, et al.A method for early detection and identification of cables based on multi-scale wavelet criterion and time-frequency feature correlation[J]. Power System Protection and Control, 2015, 43(9): 115-123.
[6] Karimi A S, Kuo C C J. A robust technique for latent fingerprint image segmentation and enhancement[C]// IEEE International Conference on Image Processing, San Diego, California, 2008: 193-200.
[7] 张子健. 面向高铁接触网缺陷检测的智能图像处理关键技术研究[D]. 杭州: 浙江大学, 2019.
[8] 戚广枫, 赵慧, 肖晓晖, 等. 高速铁路接触网吊弦动应力数值模拟及其疲劳荷载特征分析[J]. 中国机械工程, 2018, 29(9): 1063-1068.
Qi Guangfeng, Zhao Hui, Xiao Xiaohui, et al.Numerical simulation of dynamic stress and its fatigue load characteristics of catenary strings in high speed railway contact network[J]. China Mechanical Engineering, 2018, 29(9): 1063-1068.
[9] Liu Wenqiang, Liu Zhigang, Alfredo N, et al.Multi- objective performance evaluation of the detection of catenary support components using DCNNs[J]. IFAC Papers on Line, 2018, 51(9): 21-27.
[10] Huang Shize, Zhai Yachan, Zhang Miaomiao, et al.Arc detection and recognition in pantograph-catenary system based on convolutional neural network[J]. Information Sciences, 2019, 501(3): 1024-1029.
[11] Hinton G E, Osindero S, Teh Y W.A fast learning algorithm for deep belief nets[J]. Neural Computation, 2006, 18(7): 1527-1554.
[12] 范华, 任俊, 曹卫国. 一种面向低压配电网的非侵入故障定位识别方法[J]. 电气技术, 2014, 15(11): 66-69.
Fan Hua, Ren Jun, Cao Weiguo.A non-intrusive fault location and identification method for low voltage distribution network[J]. Electric Technology, 2014, 15(11): 66-69.
[13] 卢用煌, 黄山. 深度学习在身份证号码识别中的应用[J]. 应用科技, 2019, 46(1): 123-128.
Lu Yonghuang, Huang Shan.Application of deep learning in identification of ID card number[J]. Applied Technology, 2019, 46(1): 123-128.
[14] 王毅星. 基于深度学习和迁移学习的电力数据挖掘技术研究[D]. 杭州: 浙江大学, 2019.
[15] Sabour S, Frosst N, Hinton G E.Dynamic routing between capsules[J]. Neural Information Processing System, 2017, 3(1): 3856-3866.
[16] 朱娟, 陈晓. 改进胶囊网络的有序重叠手写数字识别方法[J]. 激光杂志, 2019, 40(7): 43-46.
Zhu Juan, Chen Xiao.An improved overlapping handwritten digit recognition method for capsule networks[J]. Laser Journal, 2019, 40(7): 43-46.
[17] Lin Lin, Wong Kaijuan, Arun K, et al.Evaluation of a TDMA-based energy efficient MAC protocol for multiple capsule networks[J]. EURASIP Journal on Wireless Communications and Networking, 2011, 2011(1): 121-128.
[18] Hao Chao, Liang Dong, Liu Yongli, et al.Emotion recognition from multiband EEG signals using CapsNet[J]. Sensors, 2019, 19(9): 22-31.
[19] 周本君. 基于卷积神经网络的人脸表情识别研究[D]. 南京: 南京邮电大学, 2019.
[20] 潘哲. 基于深度学习的航拍巡检图像绝缘子检测与故障识别研究[D]. 太原: 太原理工大学, 2019.
[21] 赵振兵, 徐磊, 戚银城, 等. 基于Hough检测和CV模型的航拍绝缘子自动协同分割方法[J]. 仪器仪表学报, 2016, 37(2): 395-403.
Zhao Zhenbing, Xu Lei, Qi Yincheng, et al.Automatic cooperative segmentation method of aerial insulators based on Hough detection and CV model[J]. Chinese Journal of Scientific Instrument, 2016, 37(2): 395-403.
[22] 赵海勇, 刘志镜, 张浩. 基于改进CV模型的多运动目标分割[J]. 仪器仪表学报, 2010, 31(5): 1082-1089.
Zhao Haiyong, Liu Zhijing, Zhang Hao.Multi- moving target segmentation based on improved CV model[J]. Chinese Journal of Scientific Instrument, 2010, 31(5): 1082-1089.
[23] 赖秋频, 杨军, 谭本东, 等. 基于YOLOv2网络的绝缘子自动识别与缺陷诊断模型[J]. 中国电力: 2019, 9(22): 1-10.
Lai Qiupin, Yang Jun, Tan Bendong, et al.Insulator automatic identification and defect diagnosis model based on YOLOv2 network[J]. Chinese Power, 2019, 9(22): 1-10.