摘要
对于轴承诊断的研究方法,常用的是对滚动轴承振动信号的特征提取,且单一维度信号特征提取诊断精度不高,着眼于该痛点,在多特征融合的基础上整合支持向量机(SVM)对滚动轴承进行故障诊断。方法首先对滚动轴承产生的内圈、外圈及滚动元件振动信号进行小波降噪和特征提取,筛选合适的小波基函数,继而从降噪后的振动信号中将时域、频域,以及基于集合经验模态分解方法的IMF能量特征提取,从多种维度将振动信号特征作为提取目标,避免了传统方法的信号失真及模态混叠等无效现象,并最终运用支持向量机(SVM)判别模型对整合后的特征信号进行故障诊断,诊断结果表明,相较于传统的特征提取及分类判别方法,支持向量机(SVM)依赖多维多域的融合特征集下的滚动轴承故障诊断准确率达到100%,具有很好的分类能力。
The commonly used research method for bearing diagnosis is feature extraction of vibration signals of rolling bearings,and the diagnostic accuracy of single dimensional signal feature extra ction is not high.Focusing on this pain point,the support vector machine(SVM)is integrated on the basis of multi-feat ures for rolling bearing fault diagnosis.The method firstly reduces the no ise by wavelet transform and extracts featur es from the vibration signals of the inner ring,the outer ring and the rolling element generated by rolling bearings,and th en screens the appropriate wavelet basis functions,secondly extracts the time domain,frequency domain and IMF energy features based on the ensemble empirical modal decomposition meth od from the noise reduced vibration signals.The results show that compared with the traditional feature extraction and classification discrimination methods,the accuracy of rolling bearing fault diagnosis by SVM relies on both multi-d imensional and multi-domain fusion feature sets has achieved 100%——It has good classification capability.
作者
许董浩
李程
XU Dong-hao;LI Cheng(College of Air Transportation,Shanghai Uni versity of Engineering Science,Shanghai 201620,China)
出处
《计算机仿真》
2024年第7期559-564,共6页
Computer Simulation
基金
国家社会科学基金资助项目(15BJL104)。
关键词
滚动轴承
故障诊断
小波变换
多特征
支持向量机
Rolling bearings
Fault diagnosis
Wavelet transform
Multi-features
Support vector machines
作者简介
许董浩(1997-),男(汉族),浙江省宁波市人,硕士研究生,主要研究领域为交通运输;通讯作者:李程(1980-),男(汉族),上海人,博士,教授,主要从事管理科学等方面的教学与科研工作。
