DBANet: A dual-branch dynamic convolutional temporal attention network for few-shot wind turbine bearing fault diagnosis

Key Points

• Fault diagnosis accuracy improved with the dual-branch dynamic convolutional approach, addressing few-shot learning challenges in wind turbine bearings.
• The model demonstrated a fault detection rate of up to 92% across various scenarios, validating its robustness against limited data scenarios.
• Analysis utilizes a dynamic convolutional temporal attention network to effectively learn representative features from minimal samples.
• The findings highlight the potential for enhanced reliability in real-time fault diagnosis, supporting maintenance strategies.