A Frequency-Weight Energy Operator and Empirical Wavelet Transform for Bearing Defect Detection

Abstract

To achieve the early bearing fault diagnosis, this paper proposes a novel repetitive transients detection algorithm. The Fourier spectrum of the analyzed signal is firstly calculated. Then the scale-space plane (SSP) of the Fourier spectrum is obtained by scale-space representation (SSR). With the help of the SSP, the division of the Fourier spectrum is realized. According to the division, the signal is decomposed by empirical wavelet transform (EWT). In addition, the Hilbert transform (HT) is replaced by frequency weight energy operator (FWEO) to demodulate the filtered signal. The effectiveness of the proposed method has been certified by experiment signal. The comparison with the ensemble empirical mode decomposition (EEMD) is conducted to illustrate the superior of the proposed method.