Cyclostationary Theory With Applications In Gearbox Fault Diagnosis

In consideration of the structures and specific running of the rotating machinery, vibration signal sampled usually is possessed of cyclostationary. Cyclostationary signal is a special kind of non-stationary signal, so by the stationary signal processing is not efficacious. In such a case, the application of cyclostationary theory in the gear fault diagnosis is studied in the paper with characteristics of gear vibration signal.First-order cyclic moment and cyclic autocorrelation function of simulation signal are calculated, conditions of first-order cyclostationary and second-order cyclostationary are illustrated. In cyclic autocorrelation function spectrogram the characteristic of spectral line of cyclic autocorrelation function is found to make calculation simply, analyze figure easily and provide the foundation of more application.Cyclic autocorrelation function of gear meshing simulation signal is analyzed, the result proves that the method has better de-noise performance when multiplying noise is stationary stochastic process. Simultaneously by comparison, demodulation performance of cyclic autocorrelation function is more widely than other demodulation methods, so can enhance accuracy and correctness of analyses.De-noise performance of higher-order cyclic statistics is discussed, by comparison, higher-order cyclic cumulant has better both de-noise performance and anti-interference performance, adapts to process signal of engineering. By analyzing higher-order cyclic cumulant of lag time being zero, the method is proved to possess better anti-interference performance.,moreover can analyze modulation frequency.Finally, as an example, analytical method of cyclostationary theory is applied to gearbox bench condition analysis. The result shows the system is effective and feasible.