基于ALIF和1DCNN的滚动轴承故障诊断方法  

Fault diagnosis method of rolling bearing based on ALIF and 1DCNN

作　　者：聂勇军[1] 孟金 肖英楠[3] NIE Yong-jun;MENG Jin;XIAO Ying-nan(Ddpartment of Mechanical Engineering,Guangzhou Maritime University,Guangzhou 510725,China;Sichuan Vocational and Technical College of Communications,Chengdu 611130,China;College of Engineering and Technology,Chengdu University of Technology,Leshan 614000,China)

机构地区：[1]广州航海学院机械工程系,广东广州510725 [2]四川交通职业技术学院,四川成都611130 [3]成都理工大学工程技术学院,四川乐山614000

出　　处：《机电工程》2022年第10期1390-1397,共8页Journal of Mechanical & Electrical Engineering

基　　金：国家自然科学基金资助项目(51875479);广州市教育科学规划项目(202113550)。

摘　　要：在滚动轴承的故障信号中,由于存在较多冗余信息成分的问题,会对基于一维卷积神经网络(1DCNN)的故障诊断的准确度产生干扰,为此,提出了一种基于自适应局部迭代滤波算法(ALIF)和1DCNN的滚动轴承故障诊断方法,即先对原始信号分解重构,再进行分类的智能故障诊断方法。首先,使用ALIF对原始信号进行了分解,其算法相较于其他信号分解算法有较少的模态混叠现象,这得益于保持其原始物理意义中,并最大程度地提取其表征信息,提高其故障诊断正确率;然后,使用了皮尔逊相关系数法选择与原始信号相关最大的本征模函数(IMF)进行了重构,得到了冗余信号较少的信号;最后,直接将处理后的数据作为1DCNN的输入,进行了智能故障诊断。研究结果表明:在对滚动轴承的4种故障状态进行分类的准确度方面,相较于原始方法,基于ALIF和1DCNN的方法准确度提高了8%,其分类准确度达到99%;仿真信号证明了ALIF分解性能的优越性,采用实验台采集的实际数据验证了该方法的先进性。In the fault signal of rolling bearing,there were many redundant components,which will interfere with the accuracy of fault diagnosis based on one-dimensional convolutional neural network(1 DCNN).Therefore,a rolling bearing fault diagnosis method based on adaptive local iterative filtering algorithm(ALIF)and 1 DCNN was proposed,that was,the original signal was decomposed and reconstructed before classification.Firstly,the original signal was decomposed by the ALIF,comparing with other signal decomposition algorithms,the algorithm had less modal aliasing,which was due to maintaining its original physical properties.Its characterization information was extracted to the greatest extent and the accuracy of its fault diagnosis was improved.Then,the Pearson correlation coefficient method was used to select the intrinsic mode function(IMF)that was most correlated with the original signal for reconstruction,and obtain a signal with less redundant signals.Finally,the processed data was directly used as the input of 1 DCNN for intelligent fault diagnosis.The research results show that the classification accuracy of the proposed method for the four states of the rolling bearing is improved by 8%comparing with the original method,and the classification accuracy of the proposed method reaches 99%.The superiority of the ALIF decomposition performance is proved by the simulation signal,and the advanced nature of the proposed method is verified by collecting the actual data on the experimental bench.

关 键 词：自适应局部迭代滤波 一维卷积神经网络 信号分解重构 故障分类 冗余信息成分 模态混叠 故障诊断准确率 

分 类 号：TH133.33[机械工程—机械制造及自动化] TH17