Bearing Fault Diagnosis Based On Correlation Kurtosis And Matching Tracking

With the advent of the era of the 4 industry,mechanical equipment has played a important role in modern production and life,as the part of rotating machinery,the bearing in rotating machinery also played a vital role in work can not be ignored.As the bearings are often in a complex,high load working state,it’s failure probability is much greater than other components,when the bearing not working,it will affect the production,and there is a huge security risk.Therefore,the bearing working status of the monitoring and diagnosis has a role can not be ignored.At present,the commonly method of bearing fault diagnosis is deal with the vibration signal of bearing when the bearing is working.First,the vibration signal analysis method should collected the vibration signal of bearing,and processing the vibration signal for Feature information.Finally,the fault is diagnosed based on the characteristic information.The matching pursuit algorithm(Matching Pursuit,MP)was first proposed by Mallet et al,it different from for Fourier transform,wavelet transform and Hilbert_Huang transform,the matching pursuit by creating a similar to the analysis of signal features of atoms,through continuous iteration,and selected the most similar signal characteristics the atom,finally according to the selected atom for signal reconstruction.The matching pursuit algorithm is applied in the field of mechanical fault diagnosis of mechanical fault diagnosis has made great contributions to the diagnosis of simple bearing fault,but the noise in the working environment is more complex,the fault is weak when using matching pursuit algorithm for bearing fault diagnosis effect is insufficient.In addition,the algorithm itself is a greedy iterative process,in the process of searching for the atom,the amount of calculation is huge,and to some extent,it also affects the further application of the algorithm in fault diagnosis.In this paper,the concept of correlation kurtosis is introduced to optimize the matching pursuit algorithm,and the improved algorithm is applied to the bearing fault diagnosis.This paper mainly studies the following points:(1)To decompose the signal simulation and engineering signal by EMD,and can get some IMF components,then calculate the kurtosis,and for each component of envelopment analysis,verify the sensitivity of the bearing components of kurtosis impact,and prove the sensitivity of correlated kurtosis for impact signal,and the correlated kurtosis is introduced into the matching pursuit algorithm.(2)For the matching pursuit algorithm,this paper introduces the basic theory of the algorithm,and introduces the principle of the algorithm,the construction of the atomic library and the setting of the iterative termination condition.The effectiveness of the matching pursuit algorithm for bearing fault diagnosis is verified by simulation and engineering examples.The result proved that the matching pursuit algorithm in low noise impact component is obvious,diagnosis result is good,but when the bearing fault degree of weak impact component is a lot of noise of coverage,the diagnosis results is insufficient.(3)put forward the improved kurtosis matching tracking algorithm based on matching pursuit algorithm is optimized to enhance,through the simulation analysis,engineering example analysis methods have proved the effectiveness of the proposed method,and the original matching tracking algorithm for diagnosis were compared.The comparison results show that the improved algorithm is better than the effect of diagnosis.The original algorithm.The matching pursuit algorithms are optimized by the correlation kurtosis.And The experimental and simulation results show that the improved algorithm is superior to the envelope analysis and the original matching tracking algorithm in stability and accuracy.