Research on extracting weak repetitive transients of fault rolling element bearing

The selection of the filter band for demodulation is crucial to extract the weak repetitive transients of fault rolling element bearing (REB). However, when the signal to noise is low, traditional methods for finding a suitable filter band are susceptible to the cyclostationary (CS) interference and the abnormal shock. In order to solve this problem, a new extraction method with reweighted singular value decomposition and improved Fast_kurtogram, namely RWSVD_IFK, is proposed. Firstly, by combining the new weight factors with singular components, a novel denoised strategy is presented to preprocess the original signal for reducing the CS interference. Secondly, considering that the kurtosis index is sensitive to the abnormal shock, the proposed method introduces the autocorrelation function into Fast_kurtogram to weaken the interference of the abnormal shock. And then, the squared envelope is applied to extract the weak repetitive transients from the optimal filtered signal. The simulated data and measured data of the fault REB are used to verify the validity of the proposed RWSVD_IFK. The results show that the proposed method can overcome the CS interference and the abnormal shock in finding the optimal filter band and effectively extract the weak repetitive transients of fault REB.