In order to resolve the difficulty of extracting the fault features of rolling bearings under conditions of multiple components and strong background noise, this paper proposes a rolling bearing fault feature extraction method based on modified Fourier mode decomposition (MFMD) and frequency band entropy (FBE) analysis. In order to solve the problem of the boundary frequency offset and over-decomposition of the Fourier decomposition method (FDM) under strong background noise, a method for selecting sensitive frequency bands and mode components based on band entropy and the envelope spectrum is proposed.

First, the minimum band entropy value is selected as the central frequency of the sensitive band, while the boundary of the sensitive band is determined by FBE analysis. Second, the signal is decomposed by band-limited Fourier mode decomposition in the sensitive frequency band, and several mutually orthogonal Fourier intrinsic mode functions (FIMF) and their marginal Hilbert spectra are obtained. Next, sensitive FIMFs which can reflect fault characteristics are selected according to the regional dependency relationship between the FIMFs and the FBE of the original signal. Finally, the selected FIMFs are analyzed by envelope spectrum analysis to extract the fault features.

The accurate diagnosis of bearing faults can be realized by applying this method to bearing simulation data and experimental data.

The results prove the effectiveness and superiority of the proposed method, which can provide technical support for the health evaluation of rolling bearings.