This article describes the basic theory of wavelet and wavelet analysis in signal analysis of the failure of specific applications, studied the wavelet threshold de-noising method. Elaborated the bispectrum definition, the nature, the algorithm and the physics significance, in-depth studied the bispectrum analysis in signal processing and state monitoring. For the bispectrum analysis can not suppress non-Gaussian noise as well as the shortcomings of the wavelet threshold de-noising law is no uniform standard threshold selection of insufficient, use of multi-resolution decomposition of wavelet packet technology with the bispectrum union ,given the wavelet domain bispectrum analysis method, which not only can effectively suppress noise and other components of high-frequency interference, reflect the characteristics of the fault, but also avoid wavelet noise threshold shortcomings. |