基于FRI采样的架空线路暂态电流行波故障检测  

Transient current traveling wave fault detection of overhead lines based on FRI sampling

作　　者：黄颖 徐文进 陈向叶 任晓龙 HUANG Ying;XU Wenjin;CHEN Xiangye;REN Xiaolong(State Grid Shanghai Electric Power Compangy,Shanghai 200122,China;State Grid Shanghai Pudong Electric Power Compangy,Shanghai 200120,China;State Grid Shanghai Electric Power Compangy,Shanghai 200080,China;Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100190,China)

机构地区：[1]国网上海市电力公司,上海200122 [2]国网上海浦东供电公司,上海200120 [3]国网上海市区供电公司,上海200080 [4]中国科学院计算技术研究所,北京100190

出　　处：《电子设计工程》2024年第23期94-97,共4页Electronic Design Engineering

基　　金：国网山西省电力公司研究开发项目(520530230002)。

摘　　要：为提高故障检测精度,提出基于频率响应辨识(Frequency Response Identification,FRI)采样的架空线路暂态电流行波故障检测方法。将暂态电流脉冲流拟合为高斯脉冲串,结合离散傅里叶变换和谱分析确定稀疏数据,并重构暂态电流行波信号。通过Hilbert变换获取时频谱特征,分析前三个模态分量能量,作为故障特征输入LVQ神经网络,通过迭代运算输出故障类型。实验结果表明,该方法可实现暂态电流行波信号的准确重构,故障检测成功率达到96%以上。To improve the accuracy of fault detection,a transient current traveling wave fault detection method for overhead lines based on Frequency Response Identification(FRI)sampling is proposed.Fit the transient current pulse flow into Gaussian pulse train,use FRI sampling,combine discrete Fourier transform and spectral analysis to determine sparse data,and reconstruct the transient current traveling wave signal.The time-frequency spectrum characteristics are obtained through Hilbert transform,and the energy of the first three modes is analyzed.The energy is input into the LVQ neural network as the fault characteristics,and the fault type is output through iterative operation.The experimental results show that this method can achieve accurate reconstruction of transient current traveling wave signals,and the success rate of fault detection reaches over 96%.

关 键 词：FRI采样 暂态电流行波 零化滤波器 故障特征向量 LVQ神经网络 

分 类 号：TN7[电子电信—电路与系统]