An improved Fourier decomposition method and its application in fault diagnosis of rolling bearings

In order to overcome the problem that the Fourier decomposition method can easily obtain more similar boundaries during the spectrum scanning process, resulting in too many invalid components, an improved Fourier decomposition method (IFDM) was proposed. And this method was applied to bearing fault diagnosis. First, based on the Fourier transform, several adjacent original components that are simultaneously larger or smaller than the feature mean were combined by establishing a neighborhood superposition criterion in IFDM, and a set of Fourier intrinsic mode functions (FIMF) was obtained by the method, thus reducing invalid components. Secondly, some FIMF components with kurtosis value greater than the mean value were reconstructed to extract sensitive fault feature information. Then, adaptive multi-scale weighted morphological filtering (AMWMF) was used to remove irrelevant components and background noise in the reconstructed component. Finally, the filtered signal was analyzed by spectrum. The effectiveness of the proposed method in bearing fault diagnosis was verified by the results of simulation and measured signals. At the same time, the superiority of the proposed method was verified in the comparison results with existing methods. © 2023 Chinese Vibration Engineering Society. All rights reserved.