摘要
随着用户对供电质量要求的不断提高,在线解决配电网单相接地故障定位问题成为供电部分的迫切需要。为此提出了一种中性点不接地或经消弧线圈接地的配电网单相接地故障定位的新方法,采用基于卷积型小波包能量矩(CWP-EM)的特征提取方法,对暂态电流信号进行特征向量的提取。相比传统的小波包能量特征提取方法,此方法能更有效地提取信号在各频带上的能量分布特征,并构造特征向量以作为基于免疫粒子群优化算法(IPSO)的3层小波神经网络(WNN)的训练样本集。最后,利用训练好的小波神经网络实现单相接地故障的定位。MATLAB仿真计算结果表明,提出的方法能够充分利用配电网单相接地故障信息,实现更快、更精确的单相接地故障的定位。
A new approach developed for ground-fault localization on ungrounded and high resistance grounded system is described. In the approach,an improved feature extraction method based on convolution type of wavelet packet energy moment （CWP-EM） is presented. Compared with feature extraction method based on wavelet packet energy （WPE）,the new one can better extract energy distribution feature in frequency bands,since it extracts CWP-EM and constructs CWP-EM vectors of signals,then takes the vectors as fault samples to train three-layer wavelet neural network （WNN） which is based on immune particle swarm optimization algorithm（IPSO）,finally realizes intelligent fault diagnosis. The MATLAB simulation shows that the proposed method can make full use of singlephase to ground faults information,and the fault location of this method is more accurate and reliable.
出处
《高电压技术》
EI
CAS
CSCD
北大核心
2010年第4期873-877,共5页
High Voltage Engineering
基金
国家重点基础研究发展计划(973计划)(2009CB724504)~~
关键词
故障定位
配电网
暂态电流信号
卷积型小波包能量矩
免疫粒子群优化算法
小波神经网络
fault location
distribution networks
transient current signal
convolution type of wavelet packet energy moment
immune particle swarm optimization algorithm
wavelet neural network
