Remaining Useful Life Prediction of Wind Turbine Generator Bearing Based on EMD with an Indicator

The prediction of the remaining useful life (RUL) is central to prognostics and health management (PHM) of an asset. Bearing as a critical and fragile component plays a great influence on the health condition and RUL of wind turbines. Thus it attracts more and more attentions. In this paper, a datadriven method to predict the RUL of wind turbine generator bearings is presented. Firstly, the Empirical Mode Decomposition (EMD) combined with an indicator is used to denoise and extract the bearing fault signals from raw vibration signals. Then, the fault development feathers (FDFs) are extracted from the fault signals. After that, the prediction model based on the support vector regression (SVR) is constructed to predict the RUL of the bearings. At last, the performance of the proposed method is cross-verified by actual vibration datasets from two wind turbines. The prediction result shows that the performance of the proposed method is good.