The Diagnosis of Rolling Bearing Based on the Parameters of Pulse Atoms and Degree of Cyclostationarity
Xinqing Wang,Huijie Zhu,Dong Wang,Yang Zhao,Yanfeng Li
2013-01-01
Journal of Vibroengineering
Abstract:Matching pursuit (MP) with pulse atoms adaptively matches the pulse components in the vibration signals of faulty rolling bearings, and the atomic parameters correspond with the fault features. However, the decomposition of matching pursuit with pulse atoms (MP_PA) is equivalent to searching extremum of multimodal function with considerably large calculation. Therefore, a niching particle swarm optimization, which has high global convergence ability and convergence speed, was combined with the segment and joint method to simplify MP_PA, and the efficiency was greatly improved. The parameters of pulse atom which contain almost all the information of pulse nearly completely reflect the status of bearings, by them not only the characteristic defect frequency could be extracted, but also the other features such as frequency center, damping coefficient and so on. The random in the motion of the bearings makes the pulses in vibration signals be cyclostationary rather than purely periodic, thus by using cyclostationary statistics, more accurate characteristic defect frequency could be obtained. Followed by a comparison with wavelet transform (WT) and Empirical Mode Decomposition (EMD), the techniques based on parameters of pulse atoms and degree of cyclostationarity (DCS) provide precise and explicit characteristic defect frequency of the bearing with weak defect. At last, the multi features from DCS and atomic parameters were applied to identify the status of the measured bearings accurately. These results indicate that, due to the extraction of comprehensive and accurate fault features, and strong anti-interference ability, these techniques are appropriate to fault diagnosis of bearings.