Initial Fault Time Estimation Of Rolling Element Bearing

Rolling element bearing plays a pivotal role in rotating machinery,which enable precise rotational motion and reduce rotational friction.However,bearing failure is one of the major reasons for machine breakdown.Therefore,it has received considerable attentions as it represents a frequent source of failure in the field of machine condition monitoring and fault diagnosis.However,to elicit the fault information in the early stage of rolling bearing failure remains a challenging task due to the comparatively low signal-to-noise ratio in the initial stage of failure.This makes it difficult to isolate the fault information from the original signal by using the traditional method.A signal processing approach based on the combination of EEMD,MED and Protrugram combining with back tracking strategy is proposed in this thesis to estimate the initial fault time of rolling bearing.The main contents of this thesis as follows:1.The EEMD and MED are used to de-noise the raw vibration signal of rolling bearing.The theoretical background and framework of EEMD and MED algorithms are given in this thesis.The EEMD and MED are employed to eliminate the noise in the raw vibration signal of rolling bearing.The result of the experiment showed the capability of EEMD and MED in attenuate low frequency interference and deconvolves the effect of transmission paths and highlight the periodic impulse component.2.The optimal band parameters of the output filtered signal by EEMD and MED for amplitude demodulation are selected based on Protrugram method.The raw vibration signal is filtered by EEMD and MED.The bandwidth and center frenquency for envelope analysis are optimized by Protrugram method.This method is applied to analylize the early bearing failure vibration signal.The Protrugram method proved its capacity in detecting initial faults.3.A signal processing approach based on EEMD,MED and Protrugram is proposed in this thesis to estimate initial fault time of rolling bearing.Firstly,EEMD is utilized to macerate low-frequency interference and highlight resonance component.Secondly,kurtosis index is used to reflect the degradation over entire life,and determined the fault threshold based on an abrupt increase in kurtosis value select signal(group data)earlier than the fault threshold reflected by kutosis.The MED is employed to enhance the impulse of the selected signal.The optimal band parameter for the subsequent envelope analysis is selected based on protrugram.This is followed by the comparison between the bearing characteristic frequencies with the envelope spectrum and the bearing fault type is determined.Thereafter,the initial fault time is determined by back tracking strategy.The feasibility and accuracy of the proposed approach are demonstrated by analyzing an experimental data from bearing run-to-failure test.