Research on fault diagnosis of metro traction motor bearing based on CapsNet-BIGRU and information fusion

Abstract As a critical component of the power system, the health status of subway traction motor bearings directly impacts the normal operation of the subway. However, complex and variable operating conditions, coupled with diverse types of bearing failures, increase the difficulty of fault diagnosis. To overcome the limitations of traditional feature-based diagnostic models in handling the complexities of industrial big data, this study proposes a novel fault diagnosis approach for NU216 subway traction motor bearings, which integrates CapsNet-BiGRU architecture with multi-source information fusion techniques. First, a test platform for bearings under variable operating conditions was established. Vibration and acoustic emission sensors collected signals, and a fault diagnosis dataset was constructed using feature layer fusion technology. Second, based on deep learning frameworks such as PyTorch, an improved Capsule Network (CapsNet) is integrated with a Bidirectional Gated Recurrent Unit (BIGRU) module to construct a CapsNet-BIGRU information fusion model for rolling bearing fault diagnosis, thereby improving diagnostic efficiency and accuracy. Finally, the effectiveness of the proposed method is validated through experiments involving six fault types and is compared with other mainstream approaches. Experimental results indicate that the method can effectively extract fault features under complex operating conditions and achieve high diagnostic accuracy.