自然科学版 英文版
自然科学版 英文版
自然科学版 英文版

您目前所在的位置:首页 - 期刊简介 - 详细页面

中南大学学报(自然科学版)

Journal of Central South University

第36卷    第1期    总第161期    2005年2月

[PDF全文下载]    [Flash在线阅读]

    

文章编号:1672-7207(2005)01-0097-05
支持向量机及其在机械故障诊断中的应用
何学文1,2,赵海鸣1

(1.中南大学机电工程学院,湖南长沙,410083;2.江西理工大学机电工程学院,江西赣州,341000)

摘 要: 针对目前机械故障诊断中难以获得大量故障数据样本以及特征提取和诊断知识获取困难等不足,提出了应用支持向量机进行机械故障诊断的方法,研究了将小波包分析与信号能量分解用于机械故障的特征提取。该方法将振动信号小波包分析后的信号频带能量作为特征向量,输入到由多个支持向量机构成的多故障分类器中进行故障识别和分类。该分类器只需少量训练样本,而且不必预先知道故障分类的经验知识就能实现正确分类。研究结果表明:选用不同核函数及其参数的多故障分类器对分类精度有影响;在样本不带噪声和带15%噪声情况下,支持向量机的分类精度均高于BP神经网络的分类精度,具有更好的分类性能。

 

关键字: 支持向量机;小波包分析;特征提取;故障诊断;多故障分类器

HE Xue-wen1,2, ZHAO Hai-ming1

1.School of Mechanical and Electrical Engineering, Central South University, Changsha 410083, China;2.School of Mechanical and Electrical Engineering, Jiangxi University of Science and Technology, Ganzhou 341000, China

Abstract:Aiming at the difficulty in getting adequate fault samples, extracting eigenvectors and
acquiring diagnosis knowledge in machinery fault diagnosis, a novel method for machinery fault
diagnosis based on support vector machine(SVM) was put forward, in which wavelet packet anal-
ysis and signal energy decomposition were used for the feature extraction machinery faults. According to the method, the energy of different frequency bands after wavelet packet decomposition constitutes the input vectors of support vector machine as eigenvectors, and these eigenvectors were input into multiple fault classifiers to identify faults. The new method, by which multiple faults can be diagnosed, only requires a small quantity of fault samples and it doesn’t need the field knowledge of fault diagnosis. The experimental results show that different SVMclassifiers, in which different kernel functions and their parameters are adopted, will influence the precision of fault classifiers.Under the circumstances that samples don’t include noise signal and samples include 15% noise signal,the classification precision of SVM classifiers is higher than that of BP artificial neural networks. ThusSVM classifiers show better classification performance.

 

Key words: support vector machine; wavelet packet analysis; feature extraction; fault diagnosis;
multiple fault classifiers

中南大学学报(自然科学版)
  ISSN 1672-7207
CN 43-1426/N
ZDXZAC
中南大学学报(英文版)
  ISSN 2095-2899
CN 43-1516/TB
JCSTFT
版权所有:《中南大学学报(自然科学版、英文版)》编辑部
地 址:湖南省长沙市中南大学 邮编: 410083
电 话: 0731-88879765 传真: 0731-88877727
电子邮箱:zngdxb@csu.edu.cn 湘ICP备09001153号