WT and spectral kurtosis of rolling bearing, and smooth iterative demodulation fault diagnosis method

The invention discloses a WT and spectral kurtosis of a rolling bearing, and a smooth iterative demodulation fault diagnosis method. The method comprises: firstly, decomposing an original signal by using a wavelet decomposition method; then, noise components and trend terms in a decomposition result are eliminated by utilizing rearrangement and replacement operations of the data; analyzing the signal after the first filtering by adopting a spectral kurtosis method; obtaining the center frequency and the bandwidth of an optimal filter, performing secondary filtering on a signal filtered for thefirst time by using the filter, performing envelope analysis on the signal filtered for the second time by using a smooth iterative envelope analysis method, and finally determining the fault type ofthe rolling bearing according to an envelope spectrum. The method is suitable for processing complex rolling bearing fault signals, can accurately determine the fault type of the rolling bearing, hasgood noise resistance and