Research on Identification Strategy of Fault-Sensitive Frequency for Planetary Gearboxes

Planetary gearboxes find extensive application in wind power generation, aerospace, industrial production and other fields. Harsh operational conditions and prolonged periods of continuous work often lead to various gear damages. Therefore, it is important to continuously monitor the health status of planetary gearboxes. However, the fault-sensitive frequencies of planetary gearboxes used in the existing signal models and fault diagnosis indicators are calculated theoretically or searched artificially. In response to this, this paper proposes a strategy for identifying fault-sensitive frequencies in planetary gearboxes intelligently by combining spectral amplitude analysis with three models: Random Forest (RF), Support Vector Machine (SVM) and Deep Neural Network (DNN). Experiments are carried out to verify the effectiveness and superiority of the proposed strategy.