摘要
针对过渡电阻较大造成的配电网单相接地故障选线难的问题,提出一种基于EEMD-Hilbert变换与SVM的小电流接地系统故障选线方法。首先,根据故障初相角的不同,建立两类SVM分类器,应用EEMD算法对1/6周期的暂态零序电流进行分解,通过相关系数获得零序特征电流,并对其进行Hilbert变换。然后,定义特征解析点,并分别求取虚相位特征角并构造虚相位特征角向量。最后,利用虚相位特征角向量训练与之相应的SVM分类器,输入测试集,输出分类结果。仿真结果表明,该选线方法原理简单,选线准确率高。
The single-phase ground fault of distribution network occurs with the big ground resistance, a novel fault line selection method was proposed based on EEMD-Hilbert and SVM for small current to ground system. Firstly, ac- cording to the initial phase angle after the fault, two types of support vector machine classifiers were established. The transient zero-sequence current in the first 1/6 cycle was decomposed by EEMD, and the characteristic current was obtained by calculating the correlation coefficient, which was transformed by Hilbert transform. Then, depending on the mutation of the transient zero-sequence current, the characteristic analysis points were defined. The virtual phase characteristic angle was obtained, and the virtual phase characteristic angle vectors can be calculated. Finally, the SVM classifiers were trained by using the corresponding virtual phase feature vectors. By inputting the test set, the classification results were output. The simulation results proved that the method of line selection method was simple and high accurate.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2014年第12期55-61,共7页
Proceedings of the CSU-EPSA
基金
国家自然科学基金项目(61403127)
河南省教育厅科学技术研究重点项目(12B470002
14A470004)
河南省控制工程重点学科开放实验室资助项目(KG2011-15)
河南理工大学青年基金项目(Q2012-28)
关键词
HILBERT变换
支持向量机分类器
相关系数
特征解析点
虚相位特征角向量
Hilbert transform
support vector machine （SVM） classifier
correlation coefficient
feature analysis points
virtual phase feature vectors
