Feature Frequency Extraction Method based on SAS Noise Reduction and Spectral Kurtosis

Abstract

A fault feature extraction method for rolling bearings based on sparse assisted smoothing (SAS) noise reduction and spectral kurtosis (SK) is proposed. Firstly, the simulation signal is de-noised by sparse assisted smoothing (SAS), and the performance of SAS de-noising and low-pass filtering is evaluated by RMSE index. Then, SAS de-noising and spectral kurtosis method are applied to fault diagnosis of rolling bearings, and clear characteristic frequency is extracted. Then the kurtosis index is used to compare the proposed method with the wavelet de-noising and EMD methods. The advantages of the proposed method in feature extraction are verified. The effectiveness of the proposed method is verified by simulation and data experiments.

Keywords

• Sparse assisted smoothing