Research on Fault Feature Extraction of Rolling Bearing Based on Adaptive DFIF and CYCBD

Due to the fact that the fault features of rolling bearing is easily drowned out by strong noise, it is difficult to exploit the impact components which contain fault information, a combination method based on snake optimizer (SO), direct fast iterative filtering (DFIF) and maximum second order cyclostationary blind deconvolution (CYCBD) is proposed, namely, SO-DFIF-CYCBD, for rolling bearing fault features extraction. Firstly, the fault vibration signal is adaptively decomposed employing parameter optimized DFIF to acquire several intrinsic mode function (IMF) components. Secondly, the sensitive components are screened for reconstruction based on the kurtosis-correlation coefficient criterion. Finally, the reconstructed vibration signal is deconvoluted utilizing the CYCBD method to reinforce the fault shock components, and obvious frequency components are extracted using envelope demodulation to distinguish the fault type. The reliability of the proposed approach are verified through simulation vibration signals and actual rolling bearing signals. In addition, compared to other combination model techniques in this article, the proposed approach can extract richer fault feature information and has stronger fault feature extraction capabilities.