本研究探討風力發電機齒輪箱高速軸承之故障特徵分析，研究的數據為風機高速軸承之振動量測訊號。當軸承劣化或故障，藉由加速規之量測可監測其振動量隨運轉時間之增長趨勢，經由特徵萃取可量化此軸承之劣化狀況。除考慮時域之各項特徵外，另包含頻域特徵萃取之探討，如譜峰度(spectral kurtosis，SK)、SK Mean、SK Kurtosis、邊帶功率因數(sideband power factor)等，掌握故障特徵之趨勢特性，供後續分析擇用之參考，以利進行風力發電機齒輪箱高速軸承之狀態監測與有效執行維修之排程。

This study investigates the fault feature analysis of a high-speed bearing in wind turbine gearbox. The datasets studied were obtained from the measured vibration signal of a wind turbine high-speed shaft bearing. When the bearing is degraded or faulty, the accelerometer can be used to monitor the increase of the vibration level with respect to the running time. The fault characteristics of the bearing can be quantified by feature extraction techniques. Besides the consideration of various features in the time domain, this study explores feature extraction in the frequency domain as well, such as Spectral kurtosis (SK), SK Mean, SK Kurtosis, and Sideband Power Factor, etc. The trend characteristics of fault features can be used for reference in subsequent analysis and selection, in order to facilitate condition monitoring of high-speed bearings in wind turbine gearbox and the effective execution of maintenance scheduling.