Faults Diagnosis of Rolling Bearings Based on Adaptive Fault Frequency Band Detection

The feature information is usually corrupted by the irrelevant harmonics and background noise for bearing fault signal, which makes it difficult to identify the fault symptom in time. This paper proposes a new diagnosis method to identify the incipient periodic impulsive features of rolling bearings. Firstly, the Over-Complete Rational-Dilation Wavelet Transforms (ORDWT) is employed to decompose the original fault signal to obtain several sub-bands. Secondly, a periodic impulsive index which absorbs the advantages of ACFHNR, the kurtosis and Pearson's correlation coefficient index is proposed to adaptively track the best fault frequency band. Finally, the envelop spectrum of the best fault frequency band is gained for fault diagnosis. The simulation and experiment results demonstrate that the proposed adaptive fault frequency band detection (AFFBD) method is effective.