Title: An improved gated convolutional neural network for rolling bearing fault diagnosis with imbalanced data

Authors: Changsheng Xi; Jie Yang; Xiaoxia Liang; Rahizar Bin Ramli; Shaoning Tian; Guojin Feng; Dong Zhen

Addresses: School of Mechanical Engineering, Hebei University of Technology, Tianjin 300401, China ' School of Mechanical Engineering, Hebei University of Technology, Tianjin 300401, China ' School of Mechanical Engineering, Hebei University of Technology, Tianjin 300401, China ' Department of Mechanical Engineering, Faculty of Engineering, University of Malaya, Kuala Lumpur 50603, Malaysia ' School of Mechanical Engineering, Hebei University of Technology, Tianjin 300401, China ' School of Mechanical Engineering, Hebei University of Technology, Tianjin 300401, China ' School of Mechanical Engineering, Hebei University of Technology, Tianjin 300401, China

Abstract: To improve the ability of the deep learning model to handle imbalanced data, a fault diagnosis method based on improved gated convolutional neural network (IGCNN) is proposed. Firstly, an improved gated convolution layer is proposed for feature extraction, with the batch normalisation (BN) layer applied to adjust the data distribution and enhance the generalisation performance of the model. Then, the feature learned by multiple gated convolution layers and pooling layers is fed to the fully connected layer for fault type identification. Finally, the label-distribution-aware margin (LDAM) loss function is employed to adjust the model being more sensitive to the minority class and mitigate the influence of imbalanced data on the model. Experimental validation is conducted using two bearing datasets. Results show that the proposed method is more robust than other fault diagnosis methods, with higher recognition accuracy in severely imbalanced dataset.

Keywords: rolling bearings; fault diagnosis; imbalanced data; IGCNN; label-distribution-aware margin loss.

DOI: 10.1504/IJHM.2023.130520

International Journal of Hydromechatronics, 2023 Vol.6 No.2, pp.108 - 132

Received: 30 Aug 2022
Accepted: 08 Nov 2022
Published online: 25 Apr 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article