Electrical Engineering  2020, Vol. 21 Issue (11): 32-38    DOI:
Current Issue| Next Issue| Archive| Adv Search |
Fault diagnosis method of motor bearing based on attention and multi-scale convolution neural network
Tang Si1, Chen Xinchu1, Zheng Song1,2
1. College of Electrical Engineering and Automation,Fuzhou University,Fuzhou 350116;
2. Key Laboratory of Industrial Automation Control Technology and Information Processing of Fujian Higher Education Institutes,Fuzhou 350116

Download: PDF (19116 KB)   HTML (1 KB) 
Export: BibTeX | EndNote (RIS)      
Abstract  In this paper, an improved end-to-end convolutional neural network fault diagnosis method is proposed to deal with the problem that motor fault diagnosis needs complex feature extraction process and the diagnostic accuracy is low under the variable working conditions. In this method, the fully convolutional networks was used to prevent the feature loss caused by pooling layer. At the same time, multi-scale fault features were extracted by using convolution neural network of different kernel sizes, so that the method can obtain more abundant and complementary fault feature. Then the attention mechanism was introduced to further screen the obtained fault features, highlighting the key fault features and suppressing the unimportant fault features. Finally, a series of experiments were carried out on the fault data of motor bearing. The experiment results show that the proposed method has strong fault discriminative ability and domain adaptive ability.
Key wordsconvolutional neural network (CNN)      intelligent diagnosis      attention mechanism      multi-scale feature extraction      artificial intelligence     
Received: 21 April 2020     
Service
E-mail this article
Add to my bookshelf
Add to citation manager
E-mail Alert
RSS
Articles by authors
Tang Si
Chen Xinchu
Zheng Song
Cite this article:   
Tang Si,Chen Xinchu,Zheng Song. Fault diagnosis method of motor bearing based on attention and multi-scale convolution neural network[J]. Electrical Engineering, 2020, 21(11): 32-38.
URL:  
https://dqjs.cesmedia.cn/EN/Y2020/V21/I11/32
Copyright © Electrical Engineering
Supported by: Beijing Magtech