摘要
针对变压器故障诊断分析方法预测能力不足,诊断评价准确率低的缺点,提出一种基于神经网络算法的变压器故障识别组合方法,通过对变压器绝缘油色谱中H_(2)、CH_(4)、C_(2)H_(6)、C_(2)H_(4)、C_(2)H_(2)等特征气体的检测,采用余弦相似性与随机概率构建变压器故障气体数据图结构,并将其分别输入至神经网络算法模型中,通过加权的方式将结果进行聚合。仿真结果表明:所提方法与其他神经网络方法相比,其诊断准确度得到提高。
In view of the shortcomings of the existing transformer fault diagnosis analysis methods,such as insufficient prediction ability and low accuracy of diagnostic evaluation,this paper proposes a combined method for transformer fault identification based on neural network algorithms.By detecting characteristic gases such as H_(2),CH_(4),C_(2)H_(6),C_(2)H_(4),C_(2)H_(2) in the insulating oil chromatogram of the transformer,the cosine similarity and random probability are used to construct a graph structure of the transformer fault gas data,which is then input into neural network algorithm models respectively.The results are aggregated through a weighted approach.The simulation results show that,compared with other neural network methods,the accuracy of fault diagnosis using the method proposed in this paper is improved.
作者
王钢
王琳
WANG Gang;WANG Lin(Changchun University of Technology,Changchun 130000,China)
出处
《吉林电力》
2025年第3期21-27,共7页
Jilin Electric Power
关键词
油浸式变压器
故障诊断
余弦相似性
随机概率
图卷积神经网络
oil-Immersed transformer
fault diagnosis
cosine similarity
stochastic probability
graph convolutional neural network
