Compound Fault Diagnosis of Gearbox Via Simplifying Graph Convolutional Network

Accurate detection of compound fault in gearbox is significance to ensure the reliable and safe operation of rotating machinery. Due to various unknown single or compound faults, existing methods are difficult to achieve high-precision fault diagnosis for gearbox. This paper proposes a compound fault diagnosis method based on simplifying Graph Convolutional Network (SGCN), which can detect single and compound faults in gearboxes under different operating conditions. Firstly, the frequency data of vibration signals is used to construct the undirected complete graph structure. Then, we propose the SGCN model to detect the compound faults of gears and bearings in gearbox. The Southeast University (SEU) gearbox dataset and wind turbine driven system fault simulation test bench dataset are used to validate the effectiveness of the proposed method. Compared with other two methods, the proposed method show good performance in diagnosis accuracy and efficiency.