Imbalanced fault diagnosis of rolling bearing using improved MsR-GAN and feature enhancement-driven CapsNet
- Resource Type
- ACADEMIC JOURNAL
- Liu, Jie a; Zhang, Changhe b, ⁎; Jiang, Xingxing c
- In Mechanical Systems and Signal Processing 1 April 2022 168
Highlights •An improved GAN-based DA strategy is proposed to extend imbalanced sample data.•Multiple similarity indexes are adopted to evaluate the quality of generated data.•A feature enhancement network is designed to mine fault-sensitive features.•A FED-CapsNet model is developed for imbalanced fault diagnosis.•The proposed approach performs well with extremely imbalanced bearing dataset.