Rolling Bearing Fault Diagnosis Method Based On Adaptive MCKD And VMD

As an indispensable key part of mechanical equipment,rolling bearing will have a variety of failure modes in the working process.How to accurately identify its running state is one of the key issues to ensure the normal operation of mechanical equipment.In this paper,the rolling bearing is taken as the research object,the vibration signals are measured and the fault features are extracted.The improved maximum correlation kurtosis deconvolution,maximum correlation kurtosis deconvolution and variational mode decomposition are combined.The limit learning machine based on principal component analysis is used to identify the fault modes of rolling bearing.The main work of this paper is as follows:(1)The main failure modes and causes of rolling bearing are analyzed,and the collection scheme of vibration signal of the failure bearing is designed.According to the severity,the failure modes of 6205 bearing inner and outer ring pitting and inner and outer ring crack are divided into three grades: slight,moderate and severe.The vibration signals of each fault state at 500 r / min,1000 r / min and 1500 r / min under the load of 0n,300 N and 600 N are measured,and the fault data of 12 failure modes are collected,and the preliminary time-domain and frequency-domain analysis are carried out for the signals.(2)In order to solve the problem that the parameters of traditional MCKD algorithm are strictly required and it is difficult to select parameters based on artificial experience,an improved maximum correlation kurtosis deconvolution(MCKD)algorithm is proposed.First,the optimal filter length L of the MCKD algorithm is determined by the maximum permutation entropy value,then the optimal fault period T of the MCKD algorithm is determined by the maximum kurtosis value,and the parameters selected by the artificial experience are compared and analyzed;the improved MCKD algorithm and the minimum entropy deconvolution algorithm(MED)are compared and analyzed to verify the effectiveness and correctness of the improved method.(3)In order to solve the problem that it is difficult to extract the fault features of rolling bearing in strong noise environment,an improved algorithm based on MCKD and VMD is proposed.Using the MCKD algorithm as a prefilter,the filtered signal is decomposed into VMD for the first time,and a series of IMF components are obtained.The K value of VMD decomposition is determined by the kurtosis value of IMF component and the correlation coefficient of IMF component and the signal before decomposition.VMD decomposition is carried out again,and the IMF component with larger correlation coefficient and kurtosis value is selected to reconstruct the signal,and finally the reconstructed signal is enveloped Analyze and determine the failure mode.The test results show that under the strong noise environment,the fault type of rolling bearing can be accurately determined by the method.(4)In order to improve the intelligent level of fault diagnosis of rolling bearing,a fast diagnosis method combining the limit learning machine(ELM)and principal component analysis(PCA)is proposed.Firstly,select 50 samples for each state of 13 kinds of bearing States,calculate the time-domain characteristic index of each sample to take its mean value,so that it can measure the characteristics more accurately;then input the mean characteristic value into PCA,select the main element whose cumulative contribution rate is between 80% and 90%,input it into elm,quickly carry out the fault detection of rolling bearing,and combine PCA and BP neural network,PCA and support The experimental results show that the pca-elm algorithm has faster fault diagnosis speed and higher accuracy.