Rolling bearing transfer fault diagnosis method based on adversarial variational autoencoder network
Authors:
- Yisheng ZouYZYisheng Zou
- Keming ShiKSKeming Shi
- Yongzhi LiuYLYongzhi Liu
- Guofu DingGDGuofu Ding
- Kun DingKDKun Ding
- 27 July 2021
- journal article
- research article
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 32 (11) , 115017
Abstract
The intelligent diagnosis of rolling bearing (RB) faults under different working conditions has attracted significant attention. The two main limitations of existing domain-adaptation-based fault diagnosis methods for RBs are as follows. One is that the source domain transfer fault features contain a large amount of redundant information interfering with domain adaptation. The other is that discrepancies in the distribution between the same class fault samples under different working conditions lead to low transfer diagnosis accuracy. Aiming at overcoming these two limitations, in this study, a cross-domain transfer fault diagnosis model based on Wasserstein adversarial channel compression variational autoencoder (WACCVAE) is proposed. First, fault features are channel-compressed to reduce the interference of redundant features with domain adaptation; an improved variational autoencoder network with a channel compression of fault features—WACCVAE—is proposed. Secondly, the classification module function adopts an inter-class–intra-class distance constraint to improve the distribution alignment ability of same class fault samples for different working conditions. Overall, WACCVAE can accomplish the task of cross-condition transfer fault diagnosis in RBs.Keywords
Funding Information
- National Key R&D Program of China (2020YFB1708001)
This publication has 34 references indexed in Scilit:
- A novel adaptive stochastic resonance method based on coupled bistable systems and its application in rolling bearing fault diagnosisMechanical Systems and Signal Processing, 2019
- A Hybrid Prognostics Approach for Estimating Remaining Useful Life of Rolling Element BearingsIEEE Transactions on Reliability, 2018
- Roller bearing fault diagnosis using stacked denoising autoencoder in deep learning and Gath–Geva clustering algorithm without principal component analysis and data labelApplied Soft Computing, 2018
- A new subset based deep feature learning method for intelligent fault diagnosis of bearingExpert Systems with Applications, 2018
- Deep Convolutional Transfer Learning Network: A New Method for Intelligent Fault Diagnosis of Machines With Unlabeled DataIEEE Transactions on Industrial Electronics, 2018
- Recurrent Variational Autoencoders for Learning Nonlinear Generative Models in the Presence of OutliersIEEE Journal of Selected Topics in Signal Processing, 2018
- Deep variational auto-encoders: A promising tool for dimensionality reduction and ball bearing elements fault diagnosisStructural Health Monitoring, 2018
- A novel method for intelligent fault diagnosis of rolling bearings using ensemble deep auto-encodersMechanical Systems and Signal Processing, 2018
- A New Deep Learning Model for Fault Diagnosis with Good Anti-Noise and Domain Adaptation Ability on Raw Vibration SignalsSensors, 2017
- A Discriminative Feature Learning Approach for Deep Face Recognition2016