Research on weak fault feature extraction of rolling bearings based on integrated ICEEMDAN-VME-MOMEDA algorithm

ZHAO Naizhuo; LIAN Jiapeng; GAO Yongxin

doi:10.16579/j.issn.1001.9669.2026.05.016

Abstract：[Objective]To address the challenge of accurately extracting weak fault features of rolling bearings under strong background noise,a novel fault feature extraction method integrating improved complete ensemble empirical mode decomposition with adaptive noise(ICEEMDAN),variational mode extraction(VME),and multipoint optimal minimum entropy deconvolution adjusted(MOMEDA)was proposed.The aim is to enhance the identification accuracy of weak fault characteristics.[Methods]Firstly,with minimum envelope entropy as the objective,the crested porcupine optimizer(CPO)algorithm was employed to optimize the white noise amplitude weight and the number of added noise ensembles for ICEEMDAN.The original vibration signal was decomposed,and the optimal intrinsic mode function(IMF)was selected based on the maximum envelope spectrum peak factor.Secondly,guided by minimum envelope entropy,CPO optimized the balancing parameter and center frequency of VME to extract the frequency band containing the most critical fault information from the optimal IMF.Then,targeting the maximum envelope spectrum peak factor,CPO optimized the filter length and fault pulse period for MOMEDA to enhance the fault features of the extracted signal.Finally,envelope demodulation analysis was performed on the enhanced signal to extract fault characteristic frequencies.[Results]Validated by a public dataset,the proposed method clearly extracted the inner race fault frequency(162.185 2 Hz)of the rolling bearing and its multiples,with interference frequencies effectively suppressed in the envelope spectrum.The outer race fault characteristic frequency(107.364 3 Hz)is also accurately identified.Compared to using ICEEMDAN or VME alone,this method more effectively highlights fault-induced impulses,significantly improves the signal-to-noise ratio,and provides a reliable basis for accurate fault type identification.