Research On Incipient Rolling Bearing Fault Feature Extraction Method Based On CEEMDAN

Complex equipment has been used in increasingly wide areas.Once one of its certain key component appears fault,the whole system's running would be badly influenced,which may even cause accident.Rolling bearing is one of the vulnerable parts,So the incipient rolling bearing fault feature extraction method has very important significance.Empirical mode decomposition method can adaptively decompose and process non-stationary signal,but it has mode mixing problem under the interference of noise and impact signal.Therefore Researchers proposed ensemble empirical mode decomposition method,but it also can't solve the mode mixing problem completely.So complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)be introduced.Through the research,CEEMDAN method has better inhibition to mode mixing and redundant results than EEMD.The paper proposed incipient rolling bearing fault envelope analysis method based on CEEMDAN and spectral kurtosis theory.Firstly,the analysis method decompose The original vibration signal into several components,secondly choose proper components according to its kurtosis and autocorrelation function and combine them as a new reorganized signal,thirdly calculate the spectral kurtosis figure of the reorganized signal to determine spectral position of the impact component and design a corresponding filter to purify the reorganized signal,finally implement energy operator demodulation envelopment analysis to complete the extraction of fault feature.The paper verified the effectiveness of the proposed method by processing simulation signal and experiment signal.As the CEEMDAN decomposition's effectiveness been weakened by the interference of strong noise,The paper proposed incipient fault feature extraction method based on cascade adaptive bistable stochastic resonance and CEEMDAN decomposition theory.The method firstly denoise the original vibration signal by cascade adaptive bistable stochastic resonance system,then highlight the fault signal components Through CEEMDAN decomposition.The paper verified the effectiveness of the proposed method by processing simulation signal and experiment signal.Lastly,on the basis of theoretical research,a portable vibration signal acquisition system and off-line signal analysis system software is developed by Lab VIEW and MATLAB,which put the researches into practical engineering application.