摘要
为了进一步提高基于油中溶解气分析的变压器故障诊断准确率,提出了一种基于模糊聚类和主成分分析的故障诊断方法。该方法首先利用模糊C均值聚类方法对搜集的DGA样本集进行聚类分析,将聚类中心矩阵作为标准故障谱。然后采用改进型主成分分析方法精简样本矩阵信息,通过计算样本主成分间欧拉距离确定故障类型。通过实例验证证明本文方法能有效识别故障类型,且较IEC比值法、改良三比值法等具有更高的准确率。该方法为变压器故障诊断提供了新思路,具有一定的实际应用价值。
In order to enhance fault diagnosis accuracy of transformer based on dissolved gas analysis method,a new model based on fuzzy C-means cluster and improved principal component analysis is proposed.Firstly,DGA samples collected are clustered with fuzzy C-means cluster method and cluster center matrix is used as characteristic fault spectrum.Then,improved principal component analysis method is applied to calculate main components.Finally,euclidean distance between principal component of sample matrix is employed to determined fault type.Test result shows that the proposed method achieves recognition of transformer fault effectively and with a higher accuracy than IEC ration method and improved three ration method.The presented approach provides a new way for DGA and has practical application value.
作者
覃煜
黄慧红
方健
陈雁
QIN Yu;HUANG Huihong;FANG Jian;CHEN Yan(Guangzhou Power Supply Bureau,Guangzhou 510410)
出处
《高压电器》
CAS
CSCD
北大核心
2018年第12期262-267,共6页
High Voltage Apparatus
关键词
变压器
故障诊断
油中溶解气分析
模糊聚类
主成分分析
transformer
fault diagnosis
dissolved gas analysis
fuzzy C-means clustering
principal component analysis
