Anomaly Analysis and Fault Diagnosis of Underwater Glider Based on Deep Learning Method

The extended voyage duration often exposes Autonomous Underwater Gliders (AUGs) to high pressure, low temperatures, and complex ocean currents. These factors can significantly increase the failure rates of key components in AUGs. Accurately selecting critical components and effective fault diagnosis systems is crucial for the stable operation of AUGs with complex structures in dynamic marine environments. In this paper, an anomaly analysis and fault diagnosis method are proposed by integrating FTF (Fault Tree Analysis and Failure Mode Effects and Criticality Analysis, FTA and FMECA) and hybrid depth model, and an “end-to-end” hybrid deep learning model. By integrating traditional reliability analysis techniques with deep learning algorithms, the method systematically identifies and prioritizes critical components. Experimental results show that this method achieves an accuracy rate of over 98 % in fault classification, significantly enhancing the autonomous operational capabilities of AUGs. When applied in practice, it can greatly reduce reliance on experienced operators.