Rolling bearing fault diagnosis approach using probabilistic principal component analysis denoising and cyclic bispectrum

A new approach for bearing fault diagnosis is proposed based on probabilistic principal component analysis and cyclic bispectrum with optimal cycle frequency. Generally, there are two procedures to accomplish the bearings fault diagnosis. The first one is signal denoising using probabilistic principal component analysis, which transfers the original signal to a principal component model. This procedure can keep a useful component of the signal completely and increase the signal-to-noise ratio. The second employs cyclic bispectrum to extract fault frequency. Because the cyclic frequency is a vital parameter, the optical cyclic frequency is investigated and found to be equal to the bearing center frequency. The effectiveness of the proposed method is demonstrated by numerical simulation and experimental investigation of a rolling bearing with an outer race fault. The authors' analyses also indicate that the proposed method can be used for fault diagnosis of rolling bearings.