电气工程学报 ›› 2019, Vol. 14 ›› Issue (4): 42-45.doi: 10.11985/2019.04.006

• • 上一篇    下一篇

基于决策树支持向量机算法的电力变压器故障诊断研究

孙志鹏1,孙志龙2,魏建2   

  1. 1. 东北电力大学电气工程学院 吉林 132012
    2. 国网松原供电公司 松原 138000
  • 收稿日期:2019-04-03 出版日期:2019-12-25 发布日期:2020-03-18
  • 作者简介:孙志鹏,男,1994年生,硕士研究生。主要研究方向为高压电力设备故障。E-mail:1454276041@qq.com

Research on Power Transformer Fault Diagnosis Based on Decision Tree Support Vector Machine Algorithm

Zhipeng SUN1,Zhilong SUN2,Jian WEI2   

  1. 1. College of Electrical Engineering, Northeast Electric Power University, Jilin 132012 China
    2. State Grid Songyuan Power Supply Company, Songyuan 138000 China
  • Received:2019-04-03 Online:2019-12-25 Published:2020-03-18

摘要:

支持向量机算法和决策树算法均广泛应用于电力变压器故障诊断当中。与其他分类算法相比,支持向量机的泛化性能较强,但分类效率较低,而决策树算法的有较高的分类效率。提出了一种基于分类决策树的电力变压器故障诊断模型,对于靠近决定边界的样本点用支持向量机分类,其余样本用决策树分类。经实例证明,该方法有很高的分类准确率。

关键词: 电力变压器, 故障诊断, 决策树, 多值分类, 支持向量机

Abstract:

Support vector machine algorithms and decision tree algorithms are widely used in power transformer fault diagnosis. Compared with other classification algorithms, the generalization performance of support vector machines is stronger, but the classification efficiency is lower. The decision tree algorithm has higher classification efficiency. A power transformer fault diagnosis model based on classification decision tree is proposed. The sample points close to the decision boundary are classified by the support vector machine, and the remaining samples are classified by the decision tree. The example proves that the method has high classification accuracy.

Key words: Power transformer, fault diagnosis, decision trees, multiclass classification, support vector machine

中图分类号: