Abstract: The vibration signal of the gearbox rolling bearing showed strong non-stationarity under variable speed, which made it difficult to diagnose faults. Aiming at this problem, a fault diagnosis method of rolling bearing based on speed extraction and optimized modulation signal bispectrum (MSB) was proposed. First, the synchroextracting transform method (SET) was used to extract the instantaneous speed of the reference shaft from the original vibration signal; and the feature-enhanced signal was obtained by pre-whitening and minimum entropy deconvolution (MED) filters, and it was converted into an angular domain signal combined with the extracted speed, thereby establishing the MSB in the order domain. Based on the results of MSB, an improved carrier spectrum that could better reflect the degree of nonlinear coupling of between the dominant modulation components and the carrier components was constructed. Finally, the optimal carrier slices were selected according to the improved carrier spectrum, and an improved modulation spectrum was constructed by combining MSB and bispectrum coherence function to further eliminate the interference of noise, so as to extract the significant fault characteristics of the rolling bearing. The experimental analysis results show that this method can effectively extract the fault characteristic orders of the gearbox rolling bearing under variable working conditions, and then complete the fault diagnosis of the rolling bearing, and has obvious advantages compared with the traditional methods.

Key words:  gearbox rolling bearing; fault characteristic orders; synchroextracting transform (SET); speed extraction; modulation signal bispectrum (MSB); improved modulation spectrum