高斯混合模型与小波包能量相结合的齿轮故障分类  

Gear fault classification based on Gaussian mixture model and wavelet packet energy

作　　者：肖涵[1] 李友荣[1] 吕勇[1] 

机构地区：[1]武汉科技大学机械自动化学院,222信箱430081

出　　处：《起重运输机械》2009年第1期92-95,共4页Hoisting and Conveying Machinery

基　　金：国家自然科学基金资助(50705069);湖北省自然科学基金资助(2005ABA287)

摘　　要：提出小波包能量与高斯混合模型相结合的齿轮故障分类算法。利用小波包分析提取某种模式下齿轮振动信号多层分解后的不同频带内的能量,并进行归一化处理。然后以各频带能量为元素构造该模式的特征向量,利用这些特征向量以及高斯混合模型良好的数据分布刻画能力,对该模式进行描述。最后采用贝叶斯分类器进行齿轮故障分类。采用该方法对齿轮振动信号进行故障识别,结果表明能取得比人工神经网络算法更高的识别率。The paper presents the gear fault classification based on Gaussian mixture model and wavelet packet energy, and how the energy of a specific fault mode in various frequency bands is extracted and then normalized. With the energy in various frequency bands regarded as the feature vectors of this mode, this mode is described based on GMM which is a good tool to fit the data distribution. Finally the gear faults are classified by using Bayes＇classifier. The fault recognition method proposed in this paper features higher recognition rate than artificial neutral network method.

关 键 词：故障分类 小波包能量 高斯混合模型 

分 类 号：TH132.41[机械工程—机械制造及自动化]