Identification of characteristic components in frequency domain from signal singularities

In rotating machinery condition monitoring, identification of characteristic components is fundamental in many engineering applications so as to obtain fault sensitive features for fault detection and diagnosis. This paper proposed a novel method for the identification of characteristic components in frequency domain based on singularity analysis. In this process, Lipschitz exponent function is constructed from the signal through wavelet-based singularity analysis. In order to highlight the periodic phenomena, autocorrelation transform is employed to extract the periodic exponents and Fourier transform is used to map the time-domain information into frequency domain. Case study with rolling element bearing vibration data shows that the proposed has very excellent capability for the identification of characteristic components compared with traditional methods.