| 引用本文: | 杨耿杰,许晔,高伟,洪翠,郭谋发.基于能量谱相似度自适应聚类的配电网接地故障区段定位方法[J].电力自动化设备,2021,41(3): |
| YANG Gengjie,XU Ye,GAO Wei,HONG Cui,GUO Moufa.Grounding fault section location method of distribution network based on adaptive clustering of similarities between energy spectra[J].Electric Power Automation Equipment,2021,41(3): |
|
| 摘要: |
| 分析谐振接地系统发生单相接地故障时各区段故障暂态零序电流的相似性特征，提出基于故障暂态零序电流Hilbert瞬时能量谱相似性自适应仿射传播聚类(adAP)的故障区段定位方法。所提方法对故障后首半个工频周期的故障线路各区段的故障暂态零序电流进行局部特征尺度分解(LCD)，对得到的各频带波形进行Hilbert变换得到Hilbert瞬时能量谱。计算Hilbert瞬时能量谱两两间的动态时间弯曲距离，对动态时间弯曲距离构成的相似度矩阵进行adAP，根据Silhouette指标最高的聚类情况定位故障区段。在电弧故障、存在分布式消弧线圈系统、不同消弧线圈补偿度系统、噪声干扰、两点接地、采样不同步的场景下的仿真结果和现场数据验证结果表明，所提方法能克服工程应用中可能存在的影响，鲁棒性好、准确度高。 |
| 关键词: 谐振接地系统 接地故障区段定位 暂态零序电流 局部特征尺度分解 Hilbert瞬时能量谱 动态时间弯曲距离 自适应仿射传播聚类 |
| DOI：10.16081/j.epae.202102010 |
| 分类号:TM77 |
| 基金项目:国家自然科学基金资助项目(51677030) |
|
| Grounding fault section location method of distribution network based on adaptive clustering of similarities between energy spectra |
|
YANG Gengjie1, XU Ye2, GAO Wei1, HONG Cui1, GUO Moufa1
|
|
1.College of Electrical Engineering and Automation, Fuzhou University, Fuzhou 350108, China;2.Zhangzhou Electric Power Supply Company, State Grid Fujian Electric Power Co.,Ltd.,Zhangzhou 363000, China
|
| Abstract: |
| The similarity characteristics of transient zero-sequence currents in each section of resonant grounding system during single-phase grounding fault are analyzed, based on which, a grounding fault section location method of distribution network based on adAP(adaptive Affinity Propagation clustering) of similarities between Hilbert instantaneous energy spectra is proposed. The first half power frequency cycle of the transient zero-sequence current in each section of faulty line is decomposed into waveforms of different frequency band by LCD(Local Characteristic-scale Decomposition). The Hilbert transform of the waveforms of different frequency band is carried out to obtain the Hilbert instantaneous energy spectra. The DTW(Dynamic Time Warping distance) between every two of Hilbert instantaneous energy spectra is calculated to construct the similarity matrix. Then the matrix is clustered by adAP algorithm, and the fault section is located by the largest Silhouette index. The simulation and field test results under the scenarios of arc fault, distributed arc suppression coil system, different compensation of arc suppression coil, noise interference, two-point grounding and asynchronous sampling show that the proposed method can accommodate to the underlying influencing situations in engineering application, and has good robustness and high accuracy. |
| Key words: resonant grounding system grounding fault section location transient zero-sequence current local characteristic-scale decomposition Hilbert instantaneous energy spectrum dynamic time warping distance adaptive affinity propagation clustering |
