![](http://www.meem.com.cn/static/fore_en/images/logo_01.jpg)
Founded in 1971 >
Chinese Sci-tech Core Periodicals >
British Science Abstracts (SA, INSPEC) Indexed Journals >
United States, Cambridge Scientific Abstract: Technology (CSA: T) Indexed Journals >
United States, Ulrich's Periodicals Directory(UPD)Indexed Journals >
United States, Cambridge Scientific Abstract: Natural Science (CSA: NS) Indexed Journals >
Poland ,Index of Copernicus(IC) Indexed Journals >
International Standard Serial Number:
ISSN 1001-4551
Sponsor:
Zhejiang University;
Zhejiang Machinery and Electrical Group
Edited by:
Editorial of Journal of Mechanical & Electrical Engineering
Chief Editor:
ZHAO Qun
Vice Chief Editor:
TANG ren-zhong,
LUO Xiang-yang
Tel:
86-571-87041360,87239525
Fax:
86-571-87239571
Add:
No.9 Gaoguannong,Daxue Road,Hangzhou,China
P.C:
310009
E-mail:
meem_contribute@163.com
Abstract: Aiming at the problem that the accuracy of bearing fault diagnosis decreases due to the influence of Doppler effect on train bearing signals in trackside acoustic detection, an intelligent fault diagnosis method for train bearing acoustics based on parameter-driven learning model was proposed. Firstly, it was proposed to use a kinematic parameters drive security realm model(KPD-SRM) for diagnosis in the case of unbalanced samples in the early stage, and use a kinematic parameters drive convolutional neural networks (KPDCNN) for diagnosis in the case of sample balance in the later stage. Then, in the simulation case, the proposed method was used to diagnose the bearing samples of ten different fault types with unbalanced and balanced samples respectively, and the accuracy of fault diagnosis was calculated. Finally, under the experimental conditions, the proposed method was used to diagnose the faults of the bearing samples of four different fault types with the sample unbalanced and the sample balanced respectively, and the accuracy of fault diagnosis was calculated. The research results show that the diagnostic accuracy of the simulated case is 97.5% and 96%, respectively, and the diagnostic accuracy of the experimental case is 93.5% and 97%, respectively, in the two cases of sample imbalance and sample balance. The parameterdriven learning model can effectively use historical data to improve the diagnostic accuracy without complex signal correction, and it will continue to improve with the increase of monitoring samples.
Key words: wheel pair bearing; trackside acoustic detection system(TADS); signal correction; kinematic parameters drive security realm model(KPDSRM); kinematic parameters drive convolutional neural networks (KPD-CNN); accuracy rate of fault diagnosis;sample imbalance
![](http://www.meem.com.cn/static/fore_en/images/foot_01.png)