摘要
为了应对日趋庞杂的故障监测系统数据,针对单通道信号存在的信息遗漏以及传统智能诊断手工提取特征的复杂性和不通用性,提出了全矢深度学习滚动轴承智能诊断方法。首先,用全矢谱融合双通道的振动信号,得到全矢融合后的主振矢数据,克服了单通道振动信号信息不完整的缺点;然后,在此基础上构建全矢深度神经网络,结合稀疏编码和去噪编码算法,自适应地提取故障特征。最后,使用反向传播算法进行网络参数整体微调。试验结果表明,该方法能够自适应地提取更为有效的故障特征,提高了故障诊断准确率和稳定性,改善了传统方法的复杂流程。
To handle the numerous and jumbled data from fault monitoring systems,considering information missing with the single channel signal and the complexity and non generality of traditional intelligent diagnostic manual extracting features,a method named full vector deep learning in intelligent fault diagnosis of rolling bearing is put forward. Firstly,full vector spectrum is used to fuze the binary channel signal,the main vibration vector data after full vector fusion is acquired,the disadvantage of incomplete single channel vibration signal is overcome. Then,a full-vector deep neural network is built on this basis,combining sparse coding and de-noising coding algorithm,the fault features can be extracted automatically. Finally,the back-propagation algorithm is used to fine-tune the whole network. Experimental results show that the presented method can extract more effective fault features automatically,the classification accuracy and stability of diagnosis are improved,and the complex process of traditional methods is improved.
引文
[1]温熙森.模式识别与状态监控[M].北京:科学出版社,2007:74-80.
[2] DONG S,CHEN L,TANG B,et al. Rotating machine fault diagnosis based on optimal morphological filter and local tangent space alignment[J]. Shock&Vibration,2015,2015:1-9.
[3] QIU H,LEE J,LIN J. Wavelet filter-based weak signature detection method and its application on roller bearing prognostics[J].Journal of Sound and Vibration,2006,289:1066-1090.
[4]赵学智,叶邦彦,陈统坚.基于自适应小波特征提取一体化神经网络的空调电机振动噪声识别[J].振动与冲击,2007,26(12):160-165.
[5]刘建伟,刘媛,罗雄麟,等.深度学习研究进展[J].计算机应用研究,2014,31(7):1921-1930.
[6] TRAN V T,ALTHOBIANI F,BALL A. An approach to fault diagnosis of reciprocating compressor valves using Teager-Kaiser energy operator and deep belief networks[J]. Expert Systems with Applications,2014,41(9):4113-4122.
[7]黄海波,李人宪,杨琪,等.基于DBNs的车辆悬架减振器异响鉴别方法[J].西南交通大学学报,2015,50(5):776-782.
[8] JIA F,LEI Y,LIN J,et al. Deep neural networks:A promising tool for fault characteristic mining and intelligent diagnosis of rotating machinery with massive data[J]. Mechanical Systems&Signal Processing,2016,72/73:303-315.
[9]巩晓#.基于全矢谱的非平稳故障诊断关键技术研究[D].郑州:郑州大学,2013:10.
[10] HINTON G E,SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science,2006,313:504-507.
[11] VINCENT P,LAROCHELLE H,BENGIO Y,et al. Extracting and composing robust features with denoising autoencoders[C]∥Proceedings of the 25th International Conference on Machine Learning(ICML'08),July 5-9,2008,Helsinki,Finland. New York:ACM,c2008:1096-1103.
[12] KULLBACK S,LEIBLER R A. On information and sufficiency[J].Annals of Mathematical Statistics,1951,22(1):79-86.
[13]韩捷,石来德.全矢谱技术及工程应用[M].北京:机械工业出版社,2008:65-70.
[14] LEE J,QIU H,LIN J. Rexnord technical services‘bearing data set’[Z/OL].[2018-04-10],IMS,University of Cincinnati. NASA Ames Prognostics Data Repository NASA Ames,Moffett Field,CA,USA. http:∥ti.arc.nasa.gov/project/p-rognostic-data-repository.
[15] XUE C,FENG J,GOWRISANKAR A C,et al. Practical guidelines for developing back propagation neural network models of measurement uncertainty data[J]. Journal of Manufacturing Systems,2006,25(4):239-251.
