A New System-level Fault Diagnosis Method for Train Transmission Systems Based on Graph Learning

Efficient and reliable fault diagnosis of train transmission systems is crucial for ensuring the safety of train operations. With the advancements in big data and deep learning technologies, intelligent fault diagnosis techniques have made many achievements. However, a significant limitation in existing related research is the development of separate diagnostic models for each component, which neglects the mutual coupling relationships among different components and the propagation and interaction of faults. This oversight significantly hampers the overall diagnostic accuracy of these models. To address the aforementioned problem, a novel system-level fault diagnosis method is proposed for train transmission systems. In the proposed method, convolutional neural networks are first employed to obtain low-dimensional node attributes. Then, the graph attention mechanism is introduced for feature fusion by incorporating the spatial relationships between components, enabling the identification of the health status of each node. The efficacy of the proposed method is validated by designing and carrying out fault simulation experimental of the bogie transmission system at Beijing Jiaotong University. Experimental results demonstrate a significant improvement in diagnostic accuracy for each component node, with only a minimal increase in parameters and computational complexity.