基于模型参数辨识的空间多局放源分离方法  

Method for Separating Multiple Partial Discharge Sources in Space Based on Model Parameter Identification

作　　者：刘国峰 张璨 陈大林 王徐延 郑全福 罗林根[2] Liu Guofeng;Zhang Can;Chen Dalin;Wang Xuyan;Zheng Quanfu;Luo Lingen(State Grid Jiangsu Electric Power Co.,Ltd.,Nanjing Power Supply Branch,Nanjing Jiangsu 210005,China;School of Electric Information and Electrical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]国网江苏省电力有限公司南京供电分公司,江苏南京210005 [2]上海交通大学电子信息与电气工程学院,上海200240

出　　处：《电气自动化》2024年第2期25-27,共3页Electrical Automation

基　　金：国网江苏省电力有限公司科技项目(J2021029)。

摘　　要：对多个局放源信号的有效识别与分离是局部放电检测、定位及分析的关键。为此,提出基于模型参数辨识的空间多局放源分离方法。首先利用自回归-滑动平均模型和高阶累积量参数估计方法对特高频信号进行建模;然后依据模型参数重构信号的频谱,并结合Fisher可分离度选择信号特征频率;最后基于所选择的特征频率和径向基神经网络对特高频信号类型进行分类。仿真研究验证了方法对特高频信号的频谱重构和信号分类上的有效性,在信噪比大于10 dB的未知局部放电源信号中的多个局放源分离准确性大于75%,具有较好的应用价值。The effective identification and separation of multiple partial discharge(PD)source signals is the key to PD detection,location and analysis.For this reason,a method of spatial multiple partial discharge source separation based on model parameter identification was proposed.Firstly,the ultra-high frequency(UHF)signal was modeled using autoregressive moving average model and high order cumulant parameter estimation method;then the spectrum of the signal was reconstructed according to the model parameters,and the characteristic frequency of the signal was selected based on Fisher separability;finally,UHF signal types were classified based on the selected characteristic frequency and radial basis function neural network.The simulation study verified the effectiveness of the method in spectrum reconstruction and signal classification of UHF signals.The accuracy of multiple partial discharge sources separation in unknown partial discharge source signals with SNR greater than 10 dB is more than 75%,which has good application value.

关 键 词：局部放电 多局放源 特高频 参数辨识 

分 类 号：TM835[电气工程—高电压与绝缘技术]