Contrast research of the method for the fault diagnosis of rolling bearing based on wavelet packet-bispectrum analysis and Hibert-bispectrum analysis

The fault diagnosis of rolling bearing is an important aspect of mechanical failure detect. For the sake of getting faint vibration signals of rolling bearing,the methods based on wavelet packet-bispectrum analysis and Hibert-bispectrum analysis are proposed,contrast research is done in the different circumstances. The experiment results show that they all overcome the defect of deficient fault signals in the traditional spectrum and bispectrum. The first method is superior to the that one in the exist of feeble Gaussian and non-Gaussian noise,especially in the exist of non-Gaussian noise. That one is incapable of avoiding the disturbance of non-Gaussian noise,but the first one is capable. The research of wavelet packet-bispectrum analysis method provides a exact and efficient method for the fault diagnosis of rolling bearing.