Research On Fault Diagnosis Method Of Rolling Bearing Based On IMEEMD And LSSVM

The stable operation of industrial machinery and equipment is the key factor to ensure the normal production activities of enterprises.In mechanical equipment,rolling bearing is one of the most widely used parts,but it is also the fault prone parts in mechanical equipment.If the bearing fault can not be diagnosed and repaired in time,it will not only make the enterprise unable to complete the normal production tasks,but also affect the service life of the equipment,resulting in property loss,casualties and other catastrophic consequences.Therefore,it is necessary to diagnose the fault state of rolling bearing early,and it has great practical value and theoretical research significance.At present,Empirical Mode Decomposition(EMD)algorithm and Least Square Support Vector Machine(LSSVM)algorithm are widely used in the early diagnosis of rolling bearing fault state.EMD is mainly used for feature extraction of rolling bearing faults,and LSSVM is used for classification and recognition of different bearing states.In the process of EMD decomposition,there are many problems,such as end-point effect and false Intrinsic Mode Function(IMF),which often lead to the serious discrepancy between EMD component and real component of signal.In the process of EMD decomposition,there will be modal aliasing,which will make IMF component of EMD decomposition meaningless.In addition,the super parameter selection in LSSVM has a great influence on the final classification effect,which is easy to lead to poor classification effect only through artificial selection.In view of the above problems,this thesis takes the rolling bearing of mechanical equipment motor as the research object,mainly focusing on the signal feature extraction and fault state identification method of rolling bearing fault diagnosis.On the basis of previous studies,aiming at the shortcomings of EMD algorithm,the improved algorithm(Improved Modified Ensemble Empirical Mode Decomposition algorithm,IMEEMD)and sample entropy algorithm are combined to extract the fault features of rolling bearing,and the gray wolf optimization algorithm is used to optimize the super parameters of LSSVM,and in order to improve the accuracy of fault diagnosis of rolling bearing,the research of fault diagnosis method based on IMEEMD and LSSVM was carried out.The main work is as follows:Firstly,this thesis analyzes the fault mechanism of rolling bearing.The structure classification,failure mode and vibration mechanism of the bearing are studied,and the calculation method of the natural frequency and failure frequency of each element of the bearing is briefly introduced.Secondly,in view of the non-stationary characteristics of rolling bearing fault vibration signal and the limitations of traditional time-frequency analysis methods,this thesis focuses on the research and analysis of Empirical Mode Decomposition algorithm.In view of its end effect,this thesis adopts the extremum translation method to improve the image continuation algorithm,and applies it to the improvement of Empirical Mode Decomposition algorithm The simulation results show that the improved algorithm can effectively suppress the end effect of Empirical Mode Decomposition decomposition.In order to solve the problem of modal aliasing in Empirical Mode Decomposition,the Ensemble Empirical Mode Decomposition algorithm and Complementary Ensemble Empirical Mode Decomposition algorithm are introduced.However,there are many pseudo components and large amount of computation in the process of signal decomposition of these two algorithms.In this thesis,a Modified Ensemble Empirical Mode Decomposition algorithm is adopted,which mainly includes permutation entropy threshold detection in the process of decomposition of Complementary Ensemble Empirical Mode Decomposition algorithm,and the first several modal components of abnormal signals such as intermittent signals are separated first,and the remaining signals are decomposed by empirical modal decomposition algorithm.However,the Modified Ensemble Empirical Mode Decomposition algorithm still has the problem of endpoint effect.In view of the shortcomings of MEEMD,the Improved Empirical Mode Decomposition Algorithm in this thesis is introduced into Modified Ensemble Empirical Mode Decomposition algorithm,and an Improved Modified Ensemble Empirical Mode Decomposition algorithm is proposed.Through simulation analysis and evaluation,the superiority of Improved Modified Ensemble Empirical Mode Decomposition algorithm is verified.Then,on the basis of the introduction of sample entropy algorithm,a method of fault feature extraction of rolling bearing based on the combination of Improved Modified Ensemble Empirical Mode Decomposition algorithm and sample entropy is proposed.Firstly,the signals of different states of rolling bearing are decomposed by IMEEMD algorithm,and then the correlation coefficients between each decomposition component and the original vibration signal sequence are calculated respectively.The components with the largest correlation degree with the original signal sequence are taken as the effective components,and the sample entropy is calculated as the eigenvector of vibration signal,and compared with the approximate entropy eigenvector of IMF.The analysis shows that the IMF sample entropy can better characterize the bearing fault signal characteristics,and then realize the extraction of bearing fault characteristics.Finally,aiming at the problem of classification and recognition of rolling bearing fault diagnosis,this thesis first introduces the classification principle of Square Support Vector Machine(SVM)and Least Squares Support Vector Machine(LSSVM),and the Grey Wolf Optimization algorithm(GWO)is used to optimize the penalty factor c and kernel function parameter σ,which play an important role in the classification effect of the least squares support vector machine and the GWO-LSSVM classification model is constructed.Secondly,based on the research of fault feature extraction method,this thesis proposes a fault diagnosis method of IMEEMD-GWO-LSSVM.In order to verify the superiority of this method,this method is tested and compared with other rolling bearing fault diagnosis methods.The results show that this method has the advantages of short classification time and high accuracy of fault classification,and it can more accurately classify and identify different fault states of rolling bearing.