A Novel Cyclostationarity-based Blind Demodulation Method for the Diagnosis of Bearing

Bearing is a commonly used mechanical component in the industrial field. However, the vibration signal originally collected from machine is very complicated because of the bad working environment, such as complex structure of equipment and many other interferences. Effective separation of vibration source signal is still a challenge for accurate diagnosis of the bearing. Inspired by this fact, a novel blind demodulation method is proposed in this paper to extract sensitive component for the assessment of bearing healthy status. The proposed method combines the filtering strategy of Protrugram and the indicator of second-order cyclostationarity of fault component, where a new way is introduced to determine the optimal frequency band for demodulation. The proposed method can present good robustness to some typical interferences encountered in practical scenarios. And its performance is validated by both the simulated analysis and experimental case study. The results show that the proposed method has the benefit to be applied under complex conditions.