基于组合神经网络模型的电力变压器故障诊断方法_刘娜.pdf
文本预览下载声明
18 2 2003 4
Fault Diagnosis of Power Transformer Using a
Combinatorial Neural Network
刘 娜 高文胜 谈克雄 (清华大学电机系 100084)
Liu Na Gao Wensheng Tan Kexiong
(Department of Electrical Engineering Tsinghua University 100084 China)
对故障空间的划分以及组合神经网络的构造方式, 是利用组合神经网络进 变压器故
障识别的关键在讨论变压器故障空间划分方法及其存在问题的基础上, 针对已积累的故障变压
器的大量溶解气体数据, 考察了各类故障的气体特征及聚类分析结果, 并在此基础上构造了组合
神经网络分层结构模型, 实现了对变压器故障由粗到细的逐级划分, 以提高诊断的准确性, 为制
定维修策略提供了依据最后, 结果显示了该模型的有效性
: 电力变压器 溶解气体分析 故障诊断 神经网络 聚类分析
: T 855
Abstract ethods of fault classification and organization of combinatorial neural network
(CNN) are keysto thediagnosis of power transformer faults with CNN. In this paper, based onthe
discussion of fault classification methods and a cluster analysis of dissolved gas data of thirteen usual
transformer faults, a CNN is introduced to realize the mult-i resolution recognition of the insulation
faults, which not only can make the fault diagnosis be more exace, but also is helpful to establish a
significant strategy for the repair work. Finally, the recognition results show that this model is
effective.
Ke words: Power transformer, dissolved gas analysis, fault diagnosis, neural network, cluster
analysis
,
1
1 13
, ,
[2,3]
, ,
,
( Dissolved Gas Analysis,
DGA) , , , DGA ,
[1]
8 ,
, ,
刘 娜 , 1975 , ,
高文胜 , 1968 , ,
84 20034
[4]
显示全部