Bispectrum Analysis in Fault Diagnosis of Gears

The application of bispectrum analysis in fault diagnosis of gears is studied in this paper. Bispectrum analysis is capable of removing Gaussian or symmetric non-Gaussian noise and providing more information than power spectrum analysis. The results of the research show that normal gear signals, cracked gear signals and broken gear signals can be easily distinguished by using bispectrum as the signal features. The bispectrum diagonal slice Bx(ω1,ω2) can be used to identify the gear condition automatically.