摘要
为提高电力变压器故障诊断准确率,提出基于模糊聚类和完全二叉树支持向量机的故障诊断模型,即通过模糊C均值聚类,对样本采用完全二叉树结构逐层划分,直至最后得到各故障分类。该方法克服了一般方法对故障划分不明确、分类重叠和不可分等缺点。试验表明,相比改良三比值法、支持向量机分类"一对一"和"一对多"组合,该方法在电力变压器故障诊断中具有最高的诊断准确率。
To improve the accuracy of power transformer diagnosis, the fault diagnosis model is proposed based on fuzzy clustering and complete binary tree support vector machine （SVM）. That is, through fuzzy C-means clustering, samples are divided layer by layer using complete binary tree structure until the fault classification is completed. Compared with general approaches, the method overcomes the shortcomings of unclear division and overlap classification of fault types. The method obtains the highest diagnostic accuracy among the methods mentioned in this paper.
出处
《电工技术学报》
EI
CSCD
北大核心
2016年第4期64-70,共7页
Transactions of China Electrotechnical Society
关键词
变压器油中溶解气体
模糊聚类
完全二叉树
支持向量机
Transformer dissolved gases in oil, fuzzy clustering, complete binary tree, supportvector machine
