Joint Application of Spectral Kurtosis and Velocity Synchrosqueezing Transform for Fault Diagnosis of Ball Bearing Under Nonstationary Conditions

Time-frequency method is a good tool to analysis the vibration signal for condition monitoring of rotational machinery under non-stationary conditions. However, the fault feature of bearing cannot be directly revealed using the time-frequency method, because the fault causes complex modulation of resonance frequency. To resolve this problem, this paper proposes a joint application of Spectral Kurtosis (SK) and Velocity Synchrosqueezing Transform (VST). The vibration is firstly processed by spectral kurtosis method to extract the impulse features of the signal. After this, amplitude demodulation techniques are applied to obtain the impulse envelope. Finally the VST is applied to reveal the time-frequency feature of the envelope signal. The fault, if any, can be diagnosed by identifying the uncovered signal components. The effectiveness of the proposed method is validated using experimental signals collected under non-stationary condition.