We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
基于改进 ShuffleNet V2 网络的轴承故障诊断方法.
- Authors
赵志宏; 李春秀; 杨绍普; 张然
- Abstract
Aimed at the problem that the deep neural network fault diagnosis method has too many parameters and is not applicable to mobile devices, an improved ShuffleNet V2 network fault diagnosis model is proposed. The improvement of basic unit of ShuffieNet V2 reduces the number of parameters of the model, and replaces the standard convolution with dilated convolution in the model, so as to improve the receptive field and enhance the ability of feature extraction without increasing the parameters. The CMOR wavelet is used to transform the bearing vibration signal into time —frequency spectrum and input it to improved ShuffleNet V2 network model for fault feature extraction and recognition. The test results show that the classification accuracy of improved ShuffleNet V2 network model for bearing data set from Western Reserve University is more than 99.5%, and the classification accuracy of wheel set bearing data set of railway freight cars under actual working conditions is also more than 97%. The recognition rate and generalization performance are better than other lightweight neural networks.
- Subjects
ARTIFICIAL neural networks; CASE Western Reserve University; FAULT diagnosis; FEATURE extraction; FREIGHT cars; FREQUENCY spectra; WHEELS
- Publication
Bearing, 2022, Issue 9, p70
- ISSN
1000-3762
- Publication type
Article
- DOI
10.19533/j.issn1000-3762.2022.09015