Deep learning for bearing fault diagnosis under different working loads and non-fault location point

Intelligent fault diagnosis using deep learning has achieved much success in recent years. Using deep learning method to diagnose bearing fault requires designing an appropriate neural network model and then train with a massive data. On the one hand, up to now, a variety of neural network structure...

Full description

Saved in:
Bibliographic Details
Published in:Journal of low frequency noise, vibration, and active control vibration, and active control, 2021-03, Vol.40 (1), p.588-600
Main Authors: Wang, Chongyu, Xie, Yonghui, Zhang, Di
Format: Article
Language:English
Subjects:
Deep learning
Fault diagnosis
Fault location
Neural networks
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Intelligent fault diagnosis using deep learning has achieved much success in recent years. Using deep learning method to diagnose bearing fault requires designing an appropriate neural network model and then train with a massive data. On the one hand, up to now, a variety of neural network structures have been proposed for different diagnostic tasks, but there is a lack of research of unified structure. On the other hand, the fault data of the training neural network are collected from the fault location point, which is quite different from the actual data, because the sensor cannot be located at the fault location point accurately. This paper attempts to design a unified neural network structure based on Resnet and improve the generalization performance by using transfer learning techniques. The effectiveness of the proposed method in this paper is verified using experiment under different working loads and non-fault location point.
ISSN:1461-3484
2048-4046
DOI:10.1177/1461348419889511