Xu, Yuandong, van Vuuren, Pieter A., Tan, Xiaoli, Gu, Fengshou and Ball, Andrew (2017) A Robust Method to Detect Faults of Rolling Bearings Using Ensemble Average Autocorrelation Based Stochastic Subspace Identification. In: COMADEM 2017: 30th International Congress & Exhibition on Condition Monitoring and Diagnostic Engineering Management, Monday 10th - Thursday 13th July 2017, University of Central Lancashire, Preston. (Unpublished)

Envelope analysis plays an important role in the field of bearing faults detection. Since the development of this technique, the determination of optimal bands has been a prior challenge. Fast Kurtogram (FK) is an outstanding approach to select an optimal band for further analysis; however, fast Kurtogram is not robust enough to withstand the influence of white noise and large aperiodic impulses. Hence, a more robust method is introduced to extract the narrow bands for envelope analysis, which is ensemble average autocorrelation based stochastic subspace identification (SSI). The detector performs well in denoising and highlighting the periodic impulses owing to the outstanding characteristics of autocorrelation function and stochastic subspace identification. Considering the results of simulation study and experimental evaluation, it can be concluded that the proposed method is more effective and robust to detect bearing faults than fast Kurtogram.

A robust method to detect faults of rolling bearings using ensemble average autocorrelation based stochastic subspace identification.pdf - Accepted Version
Restricted to Repository staff only

Download (616kB)