An Estimating Method For Initial Faulty Time Of Rolling Element Bearing Based On EEMD And Spectral Kurtosis

As one of the most important parts in rotating machinery, rolling bearing plays an important role in the mechanical field. It can not only sustain the load but also keep the rotation precision of the spindle and reduce the friction between the shaft and the supporting components. However, bearing failure is one of the main sources of malfunction. It have great negative impact on the performance of the whole machine. In order to isolate the faults as early as possible, the estimation of the initial faulty time of bearing becomes important and necessary.In the early period of rolling bearing failure, the ratio of noise-to-signal is high. But the fault information is often overwhelmed by the strong background noise, which make it not easy to separate the fault components from original signal contaminated by the noise by the using of traditional method. An approach based on the combination of EEMD and spectral kurtosis is proposed in this paper to predict the initial failure time of rolling bearing. The main contents of this paper as follows:(1) The EEMD is used to de-noise the vibration signal of rolling bearing. The definition and description of EEMD algorithm is given. EEMD has better effect on decomposition compared to EMD. The method based on cross-correlation coefficient and Kurtosis is adopted to de-noise the signal by the using of sample data. The experiment result shows that EEMD can weaken interference of low frequency and highlight resonance component of the high frequency effectively.(2) An optimal band-pass filter based on SK is designed to filter the vibration signal. In order to obtain a band signal which contains the abundant fault feature components, the SK filter and Fast Kurtogram which based on the pyramidal algorithm are adopted. By analyzing the simulation signal and the vibration signal of early bearing failure, it is shown that SK is efficient in detecting incipient faults and filter capacity.(3) An approach to predict the initial failure time of rolling bearing is proposed in this paper based on the combination of EEMD and spectral kurtosis. On the one hand, EEMD is used to weaken interference of the low-frequency and highlight resonance component of the high-frequency. On the other hand, spectral kurtosis (SK) is also used in this method due to its great diagnostic capacity. The energy change with time of filtered signal can be fitted and plotted out by the using of the method combined EEMD with spectral kurtosis. The energy curve is based on the historical data of the rolling bearings whole life. Then the initial failure time of rolling bearing can be estimated. The experiment results show that this method can predict the faults time earlier.