基于形态滤波及局部切空间排列算法的滚动轴承故障诊断方法研究  

Study on fault diagnosis methods of roller bearing based on morphological filtering and local tangent space alignment algorithm

作　　者：谢昌彪 韩振南[1] 黄宏臣 

机构地区：[1]太原理工大学机械工程学院,山西太原030024 [2]南车株洲电机有限公司,湖南株洲412001

出　　处：《矿山机械》2015年第5期119-124,共6页Mining & Processing Equipment

基　　金：国家自然科学基金资助项目(50775157);山西省高等学校留学回国人员科研资助项目(2011-12);山西省基础研究项目(2012011012-1)

摘　　要：机械设备中的滚动轴承具有重要作用,但当它发生微弱故障时,振动信号通常具有非线性及低信噪比的特点。此时,采用传统的线性方法进行故障诊断效果欠佳。针对这一问题,提出了将采用形态滤波技术降噪及局部切空间排列算法进行故障分类这两种方法结合使用的新思路。采用上述方法对试验中直径为0.177 8 mm的滚动轴承故障振动信号进行诊断,并将形态滤波与小波包滤波、多维尺度分析与局部切空间排列算法分别进行对比。计算结果充分证明,形态滤波降噪及局部切空间排列算法进行故障分类识别在处理早期微弱故障时具有极大优势。The roller bearing plays an important role in mechanical equipment. In the case of weak fault of the roller bearing, its vibration signals are usually characterized by nonlinearity and low ratio of signal to noise. At the moment, traditional linear methods are not effective in fault diagnosis. In view of this problem, the paper proposed a new idea of combining morphological filtering denoising technology and local tangent space alignment algorithm usually used for fault classification. The new method was applied in the test for fault diagnosis of vibration signals from defect whose diameter was 0.1778 mm on a roller bearing, and the results of morphological filtering, wavelet packet filtering, multi-dimensional scaling analysis and local tangent space alignment algorithm were compared. The comparison results thoroughly indicated the method of combining morphological filtering denoising technology and local tangent space alignment algorithm for fault classification had great advantage in dealing with early weak fault.

关 键 词：滚动轴承故障 形态滤波 流形学习法 局部切空间排列算法 模式识别 

分 类 号：TH133.33[机械工程—机械制造及自动化]