Abstract:Based on the cyclic statistics theory, the cyclo-stationary signals are processed. The second-order cyclic statistics, which are cyclic auto-correlation function and cyclic spectrum density, are mainly studied. The research shows that the cyclic frequencies with non-zero values of cyclic auto-correlation function can indicate the modulators (faults) in signals, and then it can be used as a tool of demodulation. The method of scanning cyclic frequency is proposed to search for modulators. The separated modulators are located in low frequency ranges of the cyclic domain. The results are exhibited in the three-dimensional figure of cyclic frequency, frequency and cyclic spectrum density. The demodulation is verified by the simulation signal. Applied to the diagnose faulty rolling bearing, it separated the faults of race spalling and rolling elements flaking effectively.

Second-Order Cyclic Statistics for Mechanical Fault Diagnosis