An intelligent over-sampling framework with physics-driven guidance for noise robust imbalanced rolling bearing fault diagnosis

Key Points

• Enhanced fault diagnosis achieves higher accuracy, especially under noisy conditions with imbalanced data.
• Key improvement in diagnosis accuracy reaches 92% with physics-driven guidance for noise robustness.
• Observational analysis employs an intelligent over-sampling framework to address data imbalance effectively.
• Results highlight the need for robust methods in industrial applications involving rolling bearings.