Research on feature extraction for rolling bearing fault detection in wind turbine

Feature extraction is very important in the fault detection of rolling bearing for wind turbine. More features don't mean good performance. Data analysis and experiment based on real wind turbine samples are carried out to achieve efficient fault detection. Firstly, the original signal is decomposed with improved Empirical Mode Decomposition(EMD) to get a finite number of stationary intrinsic mode functions (IMFs). Then, characteristics of amplitude domain parameters, such as mean and variance are extracted, which can be turned into a high dimensional feature matrix. Principal component analysis(PCA) is adopted to reduce the feature matrix of vibration signals from high dimension to low dimension to remove redundant information. Classification experiments show that, this method can accurately extract the effective features of the original signal, and this can reduce the overfitting phenomenon of the machine learning model, and can also improve the accuracy of fault detection.