Instantaneous modulation signal bispectrum for frequency modulation evaluation in bearing fault diagnosis

Modulation signal bispectrum (MSB) is an algorithm that assigns demodulation significance to nonlinear and nonstationary measurement signal, which provides effective technical support for frequency modulation evaluation. However, MSB is prone to spectral leakage and spurious feature peaks when analyzing measurement signals, which is attributed to sudden changes in signal truncation that cause unwanted frequency components to be generated. To address these issues, this paper proposed instantaneous modulation signal bispectrum (IMSB) for frequency modulation evaluation in bearing fault diagnosis. Firstly, measurement signal is segmented into small segments by short-time Fourier transform (STFT), and Fourier transform is utilized on the segments to obtain local information in STFT spectrum. Subsequently, the local phase information obtained from the STFT spectrum is utilized in the MSB calculation to produce the IMSB. Finally, the selected slices from IMSB are averaged to obtain IMSB detector results. The IMSB is validated by experiment signals and analyzed in comparison with state-of-the-art algorithms (e.g., (e.g., fast spectrum correlation (Fast-SC), IESCFFOgram and Autogram). Analysis results elucidate that IMSB has advantages in frequency modulation evaluation in bearing fault diagnosis.