Intelligent Fault Diagnosis of Wind Turbine Gearbox Based on Multi-stage Extreme Gradient Boosting

Abstract

Wind turbine gearbox is widely used in wind power turbine due to its excellent transmission characteristics. The quality of wind turbine gearbox has great impact on the turbine life security. With the development of monitoring technology, as a method to record the operation state of wind power turbine, time-domain and frequency-domain analysis has been mature. However, it is of great challenge for human to identify the faults, especially compound failure pattern in operating processes. At present work, a novel compound fault diagnosis method called Multi-stage extreme Gradient Boosting (MsXGB) is proposed, which can diagnose compound faults coupled with multiple individual fault simultaneously. The diagnosis results show that the test accuracy is 97%, and the train accuracy is up to 100%.