Kaiwen Xue

Rolling element bearings are fundamental to the reliability of rotating machinery, yet detecting incipient faults under heavy background noise remains a significant challenge. To address this, this paper proposes a novel diagnostic method termed modulation bispectral slice-based kurtosis (MBSK). The MBSK method begins by applying time–frequency modulation bispectrum to demodulate the vibration signal, transforming it into a two-dimensional time–frequency representation. This process highlights the coupling between resonant bands and modulation frequencies, thereby enhancing key fault signatures while suppressing noise. Subsequently, the harmonic spectral kurtosis (HSK) indicator is introduced to quantify the periodic impulse energy within each modulation frequency slice. Robust identification of the fault characteristic frequency and its harmonics is achieved by analyzing the trend of HSK versus modulation frequency. Simulation and experimental results demonstrate that MBSK effectively extracts fault features from both the outer and inner rings of rolling bearings, maintaining high diagnostic accuracy and stability under strong noise interference. Compared to conventional modulation signal bispectrum methods, MBSK exhibits superior performance in both feature extraction capability and noise immunity.