摘要
提出一种基于小波多分辨分析和支持向量机(support vector machine,SVM)的柔性直流系统单端保护方法,仅使用单端电流即可以准确区分区内、外故障,并识别直流故障类型。首先利用小波多分辨分析求得故障电流在各频带的能量百分比,构成输入特征向量;然后通过已有数据训练后的SVM模型区分并识别故障类型。结果表明,文中所提保护方法在柔性直流系统多种运行工况下均能对故障进行准确识别,且保护方法整定简单、无需数据同步和通信,具备较强的抗干扰能力。
This paper presents a single-ended protection algorithm based on wavelet and support vector machine for flexible DC systems.The algorithm can accurately detect internal faults and identify fault type using only one terminal DC current.Wavelet multi-resolution analysis are employed to extract features of internal and external faults.Percentages of transient energy in different frequency bands are used as feature vector.Then,support vector machine(SVM)is adopted to classify fault area and fault type according to the feature vector.Simulations are carried out,showing that the algorithm can identify fault section and fault type accurately under various circumstances.The method is simple and dispenses with data synchronization and communication.It also has high detection accuracy even under serious signal noise.
作者
和敬涵
罗易萍
罗国敏
张永杰
HE Jinghan;LUO Yiping;LUO Guomin;ZHANG Yongjie(School of Electric Engineering,Beijing Jiaotong University,Haidian District,Beijing100044,China)
出处
《电网技术》
EI
CSCD
北大核心
2018年第12期4022-4030,共9页
Power System Technology
基金
国家重点研发计划项目(2016YFB0900605)~~
关键词
柔性直流系统
故障识别
小波多分辨分析
支持向量机
flexible DC grids
fault identification
wavelet multiresolution analysis
support vector machine
作者简介
和敬涵(1964),女,教授,博士生导师,主要从事电力系统保护与控制、智能电网、电动汽车与V2G等方面的工作;罗易萍(1995),女,硕士研究生,通信作者,主要研究方向为直流电网保护与控制,E-mail:16121495@bjtu.edu.cn;罗国敏(1983),女,讲师,硕士生导师,主要从事电力系统故障诊断、在线监测、故障识别和定位等方面的工作;张永杰(1995),男,硕士研究生,主要研究方向为配电网保护与控制。
