A robust deep learning system for motor bearing fault detection: leveraging multiple learning strategies and a novel double loss function

Abstract

This paper proposes a robust deep learning system for motor bearing fault detection. It leverages multiple learning strategies, including transfer learning and ensemble methods, combined with a novel double loss function to enhance feature discrimination and improve classification accuracy under noisy conditions. Extensive experiments on benchmark datasets demonstrate the superiority of the proposed method over state-of-the-art approaches.

@article{tran2025robust,
  title={A robust deep learning system for motor bearing fault detection: leveraging multiple learning strategies and a novel double loss function},
  author={Tran, Khoa and Nguyen, Minh and Le, Anh},
  journal={Signal, Image and Video Processing},
  volume={19},
  number={3},
  pages={1-15},
  year={2025},
  doi={10.1007/s11760-025-03848-8}
}