Proceedings of the

The 33rd European Safety and Reliability Conference (ESREL 2023)
3 – 8 September 2023, Southampton, UK

A New Approach for Fault Diagnosis of Rolling Bearings Based on Adaptive Batch Normalization and Attention Mechanism

Jingwen Hua, Yashun Wangb and Xun Chenc

College of Intelligence Science and Technology, National University of Defense Technology, China.


This paper proposes a single branch transfer learning method with the noise reduction attention mechanism for cross-domain fault diagnosis of rolling bearing. First, adaptive batch normalization is added to the model to ensure its domain adaptation capability. Furthermore, to improve the model's ability that suppresses noise-related features in a noisy environment, the noise reduction attention mechanism is introduced. With sufficient experimental verifications carried out, the results support that our proposed method has satisfying performance.

Keywords: Rolling bearing, Intelligent fault diagnosis, Anti-noise, Deep learning.

Download PDF