摘要
现场局部放电在线监测系统所检测的原始信号一般包含多种干扰信号和不同类型的局部放电信号,不同类型的局部放电信号叠加同样也会给局部放电的诊断造成困难。聚类分析是将相似的数据对象组成多个簇的过程,通过聚类能够从大量数据中提取有价值的知识和模式,同时还可以有效地处理噪声数据。根据大量的现场测量,提取工频周期上局部放电特高频(UHF)检波信号的特征参数,采用模糊聚类的方法,排除了脉冲干扰信号。采用灰评估以及关联分析的方法,提取不同类型局部放电所对应的相位统计谱图(PRPD)的特征参数,对比实验室建立的标准局部放电类型模式库和状态模式库,智能化诊断出现场局部放电信号所表征的放电类型和放电状态。
Raw on-line monitoring partial discharge （PD） data for transformers consist of different types of PD signals as well as several kinds of interference signals. The superimposed different types of PD signals make it difficult to diagnose PD activities. Clustering analysis is a process in which the similar data objects are divided into a number of clusters. Through clustering analysis valuable knowledge and models can be extracted from a large number of data, and noise data can be deal with effectively. In this paper, the characteristic parameters of the pre-processed data in single cycle are extracted. The noise data are then eliminated with fuzzy clustering method. The classified PD signals are statistically analyzed with PRPD; PD types and PD severity are made intelligent diagnosis based on the standard models of PD type and PD state in laboratory using gray assessment method and correlation analysis.
出处
《华北电力大学学报（自然科学版）》
CAS
北大核心
2008年第6期7-12,共6页
Journal of North China Electric Power University：Natural Science Edition
关键词
局部放电
UHF检波信号
抗干扰
模糊聚类
灰评估
智能诊断
partial discharge
enveloped UHF PD signals
interference elimination
fuzzy clustering
gray assessment
intelligent diagnosis
