Independent component analysis based sub-band in time-frequency images and it's application in fault diagnosis
Hongkun Li · Zhendong yu chongji · 2010
An improved independent component analysis(ICA) method was presented for time-frequency image blind source separation(BSS).Because there are potential correlations between similar images,the efficiency of traditional ICA applied into time-frequency image BSS was not satisfied,so mutual information and kurtosis were used to study the correlation and non-Gauss behavior of image's sub-bands choosing a special sub-band as the input parameter of ICA.Through,BBS tests of simulated time-frequency images,it was shown that the new method can improve the limitation of traditional ICA.Finally,the new method was applied into the time-frequency image of a rotor's fault signals(loosing misalignment),the time-frequency images of all faults were separated successfully,so the characteristic information of each fault was obtained.