Research On The Weak Fault Diagnosis Method Of Rolling Bearing Based On MCKD And Extreme Learning Machine

With the wide use of rotating machinery equipment in industrial production,rolling bearing which is prone to fails is one of most important parts,and its operating state plays a significant role for the safety of industrial production.Because of working condition and its own structure,the vibration signal of rolling bearing is nonlinear and non-stationary and the component of failure information is often overshadowed by strong noise,which causes that it is too difficult to extract the any information from week fault characteristics.Therefore,diagnosing the weak fault of rolling bearing is of a great practical significance.The article centers on the signal feature enhancement,feature extraction,fault identification of rolling bearing and ect.The main research are demonstrates in followings:(1)With the problem of weak fault signal of rolling bearing under noise and the difficulty of extracting the fault features,the weak fault method which is based on Minimum Entropy Deconvolution(MED)and Complementary Ensemble Empirical Mode Decomposition(CEEMD)is applied in the research.Taking advantage of MED to enhance the component features of signal,using MED as the pre-filter of CEEMD decomposition to reduce the noise interruption and increase component features,diagnosing the de-noised signal with CEEMD and further exam Teager energy spectrum of its signal,with the abovementioned steps,the weak fault feature information is extracted.The effectiveness of the method is verified by simulation analysis.(2)For the sharp pulse of MED,without considering the periodicity of the impact component,the weak fault method which is based on Maximum Correlated Kurtosis Deconvolution(MCKD)and Complete Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN)is demonstrated in the research.MCKD focuses on the continuity and periodicity of the components of signal and highlights the continuous pulse component of noise interruption.With combination of CEEMDAN and MCKD achieves the extraction of weak fault characteristics of rolling bearings.Through simulation experimental analysis,this method can effectively extract the weak fault feature information of rolling bearings.(3)With the problem of identifying weak fault of rolling bearing,the weak fault method which is based on Multiscale Permutation Entropy(MPE)and Complementary Ensemble Empirical Mode Decomposition(CEEMD)is illustrated in the research.Based on extraction of weak fault features in MCKD and CEEMDAN,the method makes good use of advantages of MPE in information extraction and uses the MPE value input as a feature vector input into extreme Learning Machines to identify the bearing fault type.The simulation results show that the mentioned method shall effectively realize the diagnosis of rolling bearing fault type.