Article: MCNN-SENet: bearing fault diagnosis method based on multi-scale convolution and squeeze-and-excitation networks Journal: International Journal of Wireless and Mobile Computing (IJWMC) 2026 Vol.30 No.2 pp.159 - 167 Abstract: In modern automated mechanical systems, the health status of rolling bearings directly affects the performance and service life of the machine, which is the focus of fault diagnosis research. Traditional bearing fault diagnosis methods rely on manual feature extraction and classifier design, which have limited efficiency and accuracy. Given the problems of limited labelled samples and noise in industrial data, this study proposes a bearing fault diagnosis method based on Multi-scale Convolution Networks (MCNet) and Squeeze-and-Excitation Networks (SENet). In this method, a large convolutional kernel and multi-scale convolution are used for efficient feature extraction, and the squeeze-and-excitation blocks are combined to enhance the sensitivity and recognition ability of the network to fault features, to improve the accuracy and robustness of fault diagnosis. The experimental results show that the average accuracy of the proposed method on the bearing data set of Case Western is 99.7%, and has good performance in the noisy environment as well. Inderscience Publishers - linking academia, business and industry through research

Title: MCNN-SENet: bearing fault diagnosis method based on multi-scale convolution and squeeze-and-excitation networks

Authors: Lida Liu; Yingjie Chen; Mei Sun; Qimiao Wang; Peiguang Lin

Addresses: Tsinghua Shenzhen International Graduate School, Shenzhen, China; Shandong Runyi Intelligent Technology Co. Ltd., Jinan, China ' School of Computing and Artificial Intelligence, Shandong University of Finance and Economics, Jinan, China ' School of Finance and Taxation, Shandong University of Finance and Economics, Jinan, China ' School of Computing and Artificial Intelligence, Shandong University of Finance and Economics, Jinan, China ' School of Computing and Artificial Intelligence, Shandong University of Finance and Economics, Jinan, China

Abstract: In modern automated mechanical systems, the health status of rolling bearings directly affects the performance and service life of the machine, which is the focus of fault diagnosis research. Traditional bearing fault diagnosis methods rely on manual feature extraction and classifier design, which have limited efficiency and accuracy. Given the problems of limited labelled samples and noise in industrial data, this study proposes a bearing fault diagnosis method based on Multi-scale Convolution Networks (MCNet) and Squeeze-and-Excitation Networks (SENet). In this method, a large convolutional kernel and multi-scale convolution are used for efficient feature extraction, and the squeeze-and-excitation blocks are combined to enhance the sensitivity and recognition ability of the network to fault features, to improve the accuracy and robustness of fault diagnosis. The experimental results show that the average accuracy of the proposed method on the bearing data set of Case Western is 99.7%, and has good performance in the noisy environment as well.

Keywords: vibration signals; convolutional neural networks; attention mechanisms; bearing fault diagnosis.

DOI: 10.1504/IJWMC.2026.151587

International Journal of Wireless and Mobile Computing, 2026 Vol.30 No.2, pp.159 - 167

Received: 03 Dec 2024
Accepted: 11 Apr 2025
Published online: 09 Feb 2026 *

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

Title: MCNN-SENet: bearing fault diagnosis method based on multi-scale convolution and squeeze-and-excitation networks

Keep up-to-date