Probability warning for wind turbine gearbox incipient faults based on SCADA data

With the rapid increase in total capacity of wind turbine, condition monitoring is more essential which can efficiently guide operation and maintenance plans. The failure rate is high occurred in gearbox, while gearbox oil temperature can reflect the operating state of the transmission structure within gearbox. In this paper, fit a Support Vector Machines (SVM) regression to model gearbox oil temperature using selected variables in Supervisory Control and Data Acquisition (SCADA) data as predictors. Sequential Feature Selection (SFS) algorithm is applied to determine the number and the type of features in the feature sets. If the residual falls outside the probabilistic prediction interval, an early warning will be given in real time. It is verified that the method proposed can give an early warning about 10 days before the actual faults.