Research on Intelligent Patrol System Based on Multi-Perception Technology

doi:10.16628/j.cnki.2095-8188.2021.12.009

Abstract

Abstract:

In order to ensure the safe and reliable operation of relay protection device,it is necessary to inspect the protection device regularly.In this paper,an intelligent inspection system based on multi perception technology is developed.With the help of the intelligent cabinet composed of high-definition and infrared cameras,temperature and humidity sensors,smoke sensors and other multi-perception devices installed in the front and back of the cabinet,the automatic inspection of the protection device is realized through image recognition technology and intelligent inspection and fault diagnosis technology.Firstly,the overall architecture and data flow diagram of the intelligent patrol system are introduced.Then,it focuses on the image recognition technology based on convolution neural network,which is a key technology in the inspection system.Through the construction of neural networks for different inspection items such as device appearance recognition and pressing plate recognition,the abnormal equipment and circuit can be identified.The recognition accuracy of all items is above 96%,and the effect is good.Furthermore,it introduces another key technology in the inspection systemintelligent inspection and fault diagnosis technology,introduces its working principle,and gives an example of inspection and fault diagnosis.The intelligent inspection system based on multi-sensor technology lays a foundation for the realization of unmanned or few people inspection of relay protection.

Key words: relay protection equipment, multi-perception, intelligent cabinet, inspection, monitoring

CLC Number:

TM77

CUI Yu, WANG Ye. Research on Intelligent Patrol System Based on Multi-Perception Technology[J]. LOW VOLTAGE APPARATUS, 2021, 0(12): 52-57.

Figures/Tables 7

References 20

[1]	陈国平, 王德林, 裘愉涛, 等. 继电保护面临的挑战与展望[J]. 电力系统自动化, 2017, 41(16):1-11,26.
[2]	LAAKSONEN H J. Protection principles for future microgrids[J]. IEEE Transactions on Power Electronics, 2010, 25(12):2910-2918. doi: 10.1109/TPEL.2010.2066990
[3]	张烈, 王德林, 刘亚东, 等. 国家电网 220 kV 及以上交流保护十年运行分析[J]. 电网技术, 2017, 41(5):1654-1659.
[4]	智能变电站保护设备在线监视与诊断装置技术规范:Q/G DW 11361—2017[S].
[5]	智能变电站继电保护在线监视和智能诊断技术导则:DL/T 1663—2016[S].
[6]	张启芳, 杨洪山, 郭乃网, 等. 基于大数据的电力设备故障诊断与预测研究及应用[C]// 2017智能电网新技术发展与应用研讨会, 2017:392-395.
[7]	笃峻, 叶翔, 葛立青, 等. 智能变电站继电保护在线运维系统关键技术的研究及实现[J]. 电力自动化设备, 2016, 36(7):163-168,175.
[8]	裘愉涛, 周震宇, 杨剑友, 等. 继电保护远程运维技术研究与应用[J]. 电力系统保护与控制, 2018, 46(18):17-24.
[9]	曾治安, 姚树友, 郑晓玲, 等. 基于移动互联网技术的继电保护设备智能运维管理模式探讨[J]. 电力系统保护与控制, 2019, 47(16):80-86.
[10]	张华宁, 郑少明, 王帆, 等. 继电保护在线监视与分析系统的功能提升与实践[J]. 电气应用, 2015, 34(S2):469-474.
[11]	王跃强, 廖华兴, 袁晓青, 等. 基于保信系统的继电保护状态评价系统的开发与应用[J]. 电力系统保护与控制, 2014, 42(11):134-139.
[12]	AZIMI-SADJADI M R, CITRIN S, SHEEDVASH S. Supervised learning process of multi-layer perceptron neural Networks using fast least squares[C]// Acoustics,Speech,and Signal Processing,1990.ICASSP-90,1990 International Conference on.IEEE, 1990:1381-1384.
[13]	DONG Y, HINTON G, MORGAN N, et al. Introduction to the special section on deep learning for speech and language processing[J]. IEEE Transactions on Audio,Speech,and Language Processing, 2012, 20(1):4-6. doi: 10.1109/TASL.2011.2173371
[14]	ANDERSEN L N, LARSEN J, HANSEN L K, et al. Adaptive regularization of neural classifiers[C]// Neural Networks for Signal Processing[1997]VII.Proceedings of the 1997 IEEE Workshop,IEEE, 1997:24-33.
[15]	徐舒玮, 邱才明, 张东霞. 基于深度学习的输电线路故障类型辨识[J]. 中国电机工程学报, 2019, 39(1):67-76.
[16]	杨为, 朱太云, 张国宝, 等. 电力物联网下基于卷积神经网络和迁移学习的GIS局部放电模式识别分类方法研究[J]. 高压电器, 2020, 56(9):20-25,32.
[17]	GUYON I M, LECUN Y A, ALBRECHT P, et al. Design of a neural network character recognizer for a touch terminal[J]. Pattern Recognition, 1991, 24(2):105-119. doi: 10.1016/0031-3203(91)90081-F
[18]	孙汉文, 李喆, 盛戈皞, 等. 基于机器学习与卷积神经网络的放电声音识别研究[J]. 高压电器, 2020, 56(9):107-113.
[19]	NOORI M R, SHAHRTASH S M. Combined fault detector and faulted phase selector for transmission lines based on adaptive cumulative sum method[J]. IEEE Transactions on Power Delivery, 2013, 28(3):1779-1787. doi: 10.1109/TPWRD.2013.2261563
[20]	苏晓, 柳明. 串补装置继电保护测试平台设计与开发[J]. 自动化仪表, 2020, 41(4):78-83,88.

位置	巡视内容
屏前	装置指示灯、液晶屏等外观是否异常	开关把手位置是否正确	压板位置是否正确
屏后	端子排接线是否规范	接地铜牌是否脱落	电缆是否破损
	光纤是否弯折	备用光纤是否无防尘帽	屏柜封堵是否正常
	屏柜内是否有异物

位置	项目名称	网络结构	卷积核	批处理数量	训练次数/次	正确率/%
屏前	装置异常识别神经网络	5C-2S-8C-2S-3C-2S	4、3、2	50	200	97.43
	开关把手状态识别神经网络	6C-2S-10C-2S-3C-2S	3、4、2	50	200	96.48
	压板状态识别神经网络	8C-2S-16C-2S-16C-2S	3、3、2	50	200	98.05
屏后	端子排接线识别神经网络	4C-2S-8C-2S-3C-2S	3、2、2	50	200	97.22
	接地铜牌识别神经网络	3C-2S-8C-2S-3C-2S	4、4、2	50	200	96.32
	电缆破损识别神经网络	4C-2S-4C-2S-2C-2S	3、4、3	50	200	98.12
	光纤折断识别神经网络	7C-2S-5C-2S-2C-2S	3、2、2	50	200	96.58
	备用光纤防尘帽识别神经网络	6C-2S-6C-2S-3C-2S	4、3、2	50	200	96.77
	屏柜封堵识别神经网络	5C-2S-4C-2S-3C-2S	3、3、3	50	200	97.25
	屏柜异物识别神经网络	6C-2S-8C-2S-2C-2S	4、3、3	50	200	97.51