Fault Diagnosis Of Rolling Bearing Based On Teager Energy Operator

As the main transmission parts of mechanical equipment,rolling bearings have got considerable attention from researchers and enterprises.However,influenced by many factors,rolling bearings are prone to failure.Rolling bearings can bear load and transfer movement,once it damaged,it will result in huge economic losses.While the disaster can be minimized if the mechanical equipment be detected and repaired in the early stage.To this end,the bearing fault detection research has been proposed in this paper.Generally,the vibration signal generated by the bearing contains a lot of external noise,and the slight fault impact component is submerged.How to eliminate the interference noise from the complicated signal so as to highlight the fault feature is the key to diagnosing the bearing.In this paper,energy operator demodulation is used to diagnose the fault,the main contents are as follows:(1)First of all,the generalized detection filter and Hilbert demodulation are introduced,and their demodulation effects are verified respectively by reliable data.The results show that although both have certain value,the former has a very slow demodulation speed and frequency aliasing,the latter is prone to signal modulation and end effect.(2)To solve the above problems,Teager energy operator demodulation is used to handle the wheel bearing sound signal and compared it with Hilbert.The results show that the energy operator demodulation is more superior than Hilbert demodulation.However,the fatal flaw of energy operator is sensitive to noise,it is of great difficulty for it to diagnose low signal-to-noise ratio vibration signal.In this case,the diagnostic methods of MED(minimal entropy decovolution)-Teager energy operator and SK(spectral kurtosis)-Teager energy operator are proposed one after another.Firstly,the noise signal is filtered by MED and SK respectively,then the signal is processed by energy operator for a better feature enhancement.The effectiveness of the proposed methods has been proved by experimental data.(3)Although MED and SK can eliminate most of the noise,some in-band noise remains.In this paper,1.5-dimensional Teager energy spectrum and the frequency-weighted energy operator are implied to solve the problem.Ultimately,proposing two diagnostic methods,they are 1.5-dimensional Teager energy spectrum combined with wavelet packet decomposition and frequency-weighted energy operator fused with 1.5-dimensional spectrum.The validity and practicability of the above two methods are verified by a variety of data.