基于决策树支持向量机模型的变压器故障诊断法  

Fault diagnosis of power transformer using SVM of decision tree

作　　者：谢红玲[1] 刘新元[1] 李燕青[1] 

机构地区：[1]华北电力大学电气与电子工程学院,河北保定071003

出　　处：《华北电力大学学报（自然科学版）》2010年第6期29-33,共5页Journal of North China Electric Power University：Natural Science Edition

摘　　要：根据对变压器常见故障原因和危害程度的分析,构造了电力变压器故障诊断决策树。其判定级别为自上而下,而决策树的每一叶都对应着一种具体的故障模式,并选用不同的支持向量机单元模块作为决策树不同分支中的基本分类器,建立组合支持向量机模型,实现了对故障的多分辨识别,该方法提高了故障分析及预测的准确度。应用结果表明该系统模型是富有成效的。A decision tree method is presented based on the synthetic analysis of the cause and injury extent of transformer malfunction.The judging grade is directed top-down with each leaf serving as a concert fault mode.While each branch corresponds to different SVM,a multi-resolution identification of transformer malfunction can thus be constructed.This method highly improved the correctness of the fault diagnosis,and the application results verified the effectiveness of this system model.

关 键 词：电力变压器 溶解气体分析 支持向量机 故障诊断 决策树 

分 类 号：TM407[电气工程—电器]