• Jan 04, 2024 News!IJFCC will adopt Article-by-Article Work Flow
  • Jun 03, 2024 News!Vol.13, No.2 has been published with online version.   [Click]
  • Dec 05, 2023 News!Vol.12, No.4 has been published with online version.   [Click]
General Information

Prof. Pascal Lorenz
University of Haute Alsace, France
It is my honor to be the editor-in-chief of IJFCC. The journal publishes good papers in the field of future computer and communication. Hopefully, IJFCC will become a recognized journal among the readers in the filed of future computer and communication.

IJFCC 2024 Vol.13(1): 13-22
DOI: 10.18178/ijfcc.2024.13.1.612

Machine Fault Diagnosis Based on a Multi-Input Multi- Branch Deep Learning Network with Attention Mechanisms

Lianghui Zou1, Haoxin Qin2, Qidong Lu3, and Zhiliang Qin 1,4, *
1. Weihai Beiyang Electrical Group Co., Ltd, Weihai, Shandong, China
2. Weihai Ivy Foreign Language School, Weihai, Shandong, China
3. Weihai Research Institute of Industry Technology, Shandong University, China
4. School of Mechanical, Electrical and Information Engineering, Shandong University, China
Email: zoulianghui@beiyang.com (L.H.Z.); qinhaoxin73@gmail.com (H.X.Q.); lqd19922012@163.com (Q.D.L.); qinzhiliang@beiyang.com (Z.L.Q.)
*Corresponding author

Manuscript received December 12, 2023; revised January 12, 2024; accepted January 30, 2024; published March 22, 2024

Abstract—In this paper, we propose a multi-input attentionbased deep-learning architecture for machine fault diagnosis, which is a challenging task due to environmental noises and signal interference that are inevitable in practical industry environments. Specifically, we develop an attention mechanism to focus on the most effective signal characteristics and achieve highly accurate fault classifications under various working conditions. First, we construct multi-dimensional features to characterize both time-domain and frequency-domain properties of machine vibration signals. Afterwards, we design a novel multi-input multi-branch (MIMB) architecture incorporating multiple sub-networks to enhance the learning of discriminant capabilities. The derived features from each subnetwork are fused to form an input to the final classification layer. To verify the effectiveness of the proposed approach, we conduct comprehensive experiments on the bearing database of Universität Paderborn, Germany, which is generally recognized as the benchmark to compare the performance of various algorithms. Numerical results show that the proposed approach achieves the state-of-the-art classification accuracy and has a noticeable performance gain over the previous schemes in the literature.

Index Terms—fault diagnosis, feature extraction, attention mechanism, multi-input network


Cite: Lianghui Zou, Haoxin Qin, Qidong Lu, and Zhiliang Qin, "Machine Fault Diagnosis Based on a Multi-Input Multi- Branch Deep Learning Network with Attention Mechanisms," International Journal of Future Computer and Communication, vol. 13, no. 1, pp. 13-22, 2024.

Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0)

Copyright © 2008-2024. International Journal of Future Computer and Communication. All rights reserved.
E-mail: ijfcc@ejournal.net