[1]高金峰,秦瑜瑞,殷红德.基于小波包变换和支持向量机的故障选线方法[J].郑州大学学报(工学版),2020,41(01):63-69.[doi:10.13705/j.issn.1671-6833.2020.01.003]
 Gao Jinfeng,Qin Yurui,Yin Hongde.Fault Line Selection Based on Wavelet Packet Transform and Support Vector Machine[J].Journal of Zhengzhou University (Engineering Science),2020,41(01):63-69.[doi:10.13705/j.issn.1671-6833.2020.01.003]
点击复制

基于小波包变换和支持向量机的故障选线方法()
分享到:

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
41
期数:
2020年01期
页码:
63-69
栏目:
出版日期:
2020-03-10

文章信息/Info

Title:
Fault Line Selection Based on Wavelet Packet Transform and Support Vector Machine
作者:
高金峰秦瑜瑞殷红德
1. 郑州大学电气工程学院;2. 国网平顶山供电公司
Author(s):
Gao Jinfeng 1Qin Yurui 1Yin Hongde 2
1. School of Electrical Engineering, Zhengzhou University; 2. State Grid Pingdingshan Power Supply Company
关键词:
故障选线小波包变换 多分类支持向量机
Keywords:
fault line selectionwavelet packet transformmulti-classificationsupport vector machine
DOI:
10.13705/j.issn.1671-6833.2020.01.003
文献标志码:
A
摘要:
配电网发生单相接地故障时,故障线路与健全线路的零序电流非工频分量差异明显。针对在高阻接地时,以模极大值极性为判据导致选线成功率不高的问题,文中给出了一种利用零序电流非工频分量和支持向量机相结合的选线方法。该方法通过小波包变换分解各线路零序电流,按能量最大原则选取特征频带,将不同线路在特征频带上的能量与模极大值作为特征向量,以故障线路标号为分类目标,把故障选线转化为多分类问题,使用支持向量机预测故障线路。通过大量仿真得到训练样本,利用K折交叉验证和网格搜索对支持向量机进行参数寻优。在测试集上的结果表明,该方法准确,可靠。在不同接地距离、接地电阻、故障初始相角下均能正确选线。
Abstract:
When single-phase grounding fault occurs in distribution network, the non-power frequency components of zero-sequence current between fault line and sound line are obviously different. In order to solve the problem that the success rate of line selection is not high when the modulus maxima polarity is used as criterion in high resistance grounding, a method of line selection based on zero sequence current non-power frequency component and support vector machine is presented in this paper. This method decomposes the zero sequence current of each line by wavelet packet transform, chooses the characteristic frequency band according to the principle of maximum energy, takes the energy and modulus maxima of different lines in the characteristic frequency band as the characteristic vector, takes the fault line label as the classification target, transforms the fault line selection int o multi-classification, and uses support vector machine to predict the fault line. A large number of training samples are obtained through simulation. K-fold cross-validation and grid search are used to optimize the param eters of support vector machine. The results on the test dataset show that the method is accurate and reliable. The correct line selection can be achieved at different grounding distances, grounding resistances and initial phase angles of faults.
更新日期/Last Update: 2020-02-22