摘要
直流电弧没有过零点特性,难以自行熄灭,易引发电气火灾,威胁直流配用电安全。通过搭建电弧模拟发生实验平台,采集不同电压下电弧信号,研究电弧电阻、电压等特性,分析了不同电压下电弧特征的差异性。以GB/T 35727—2017中低压直流配网标称电压优选值为依据,研究220 V、750 V、1 500 V3种不同电压等级电弧检测特征,对比傅里叶、小波、奇异值分解等频域特性,最大值、最小值、平均值、标准差等时域特性,得到不同电压等级下电弧检测的优质判据,证明了不同时频域特征适用于不同的电压等级。结合神经网络模型,设计了适用于750 V系统的电弧检测算法,算法检测精度为99.12%。
With no zero-crossing characteristic,DC arc is difficult to extinguish by itself,and is inclined to cause electrical fire,threatening the safety of DC power distribution.The differences of arc characteristics under different voltages are analyzed by building an experimental platform for arc simulation,collecting arc signals under different voltages,and studying the characteristics of arc resistance and voltage.Based on the preferred value of the nominal voltage of the medium and low voltage DC distribution network in GB/T 35727—2017,the differences of the detected arc characteristics at three different voltage levels of 220 V,750 V and 1500 V are studied,with the frequency domain characteristics such as Fourier transform,wavelet transform and singular value decomposition,as well as the time domain characteristics such as maximum,minimum,average and standard deviation compared,to obtain high-quality criteria for arc detection at different voltage levels,which proves that the frequency domain characteristics at different times are applicable to varied voltage levels.Combined with the neural network model,the arc detection algorithm suitable for 750 V system is designed,and the detection accuracy of the algorithm is 99.12%.
作者
刘国强
李国春
郝亚楠
李贵海
吴中杰
LIU Guoqiang;LI Guochun;HAO Yanan;LI Guihai;WU Zhongjie(State Grid Shandong Electric Power Research Institute,Jinan 250003,China)
出处
《中国电力》
CSCD
北大核心
2024年第2期94-102,共9页
Electric Power
基金
国家电网有限公司科技项目(输变电设备早期火灾探测及抑制关键技术研究
2021A-077)。
关键词
直流配电网
电弧
小波包
奇异值
时频域特性
DC distribution network
electric arc
wavelet packet
singular value
time-frequency domain characteristics
作者简介
通信作者:刘国强(1976-),男,硕士,高级工程师,从事电力火灾消防技术及电力水处理研究,E-mail:liuguoqiang1010@163.com。
