Research In Locomotive Bearing Fault Diagnosis Method Based On Variational Mode Decomposition

With the development of railway transportation system,the running speed and carrying capacity of locomotives are gradually increasing,and the running safety of locomotives is becoming more and more important.As the key moving parts of locomotives,the working performance of bearings will directly affect the operation of locomotives as a whole.Therefore,fault monitoring and diagnosis of locomotive bearings is of great significance to guarantee the safe and reliable operation of locomotives.Aiming at the feature extraction of locomotive bearing,this paper studies the fault diagnosis method of locomotive bearing based on variational mode decomposition,and the research work is as follows:In order to extract the effective modal component of the signal,the traditional mode decomposition method is studied.The results show that the empirical mode decomposition(EMD)method has good reconstruction and orthogonality,but there exists the phenomenon of mode mixing;and the ensemble empirical mode decomposition(EEMD)method eliminates the pattern aliasing by adding the white noise,however,the difference of Gaussian white noise will affect the number of modal components,poor reconfiguration,long running time and low efficiency,and the EMD and EEMD methods lack rigorous analytic expression in mathematics.Variational mode decomposition(VMD)method overcomes the disadvantages of traditional mode decomposition and has better separation characteristics in frequency domain.In this paper,the principle and algorithm of variational mode decomposition are studied.It determines the central frequency and bandwidth of each modal component by iteratively searching the optimal solution of variational model.The experimental results show that the selection of VMD modes and penalty factors will affect the result of VMD decomposition: When the number of modes is small,the correlation between the modal components and the original signals is relatively large,which contains more information components,but the decomposition accuracy is reduced,and when the number of modes is large,the modal components contain less information components,and prone to excessive decomposition phenomenon;when the penalty factor is too small,the bandwidth of each modal component is higher,the center frequency overlap phenomenon is easy to appear,the decomposition accuracy is reduced,the bandwidth of each modal component decreases with the increase of penalty factor,and the decomposition accuracy is improved.With the correlation of the modal component and the original signal as the index,the ideal mode number can be determined,and the sampling frequency as the penalty factor can improve the decomposition accuracy.In order to identify the fault type of locomotive rolling bearing,a feature extraction and analysis method based on VMD is proposed.Firstly,according to the characteristics of locomotive structure,the modal number is selected rationally,then the vibration signal is decomposed by VMD,and the characteristic frequency is determined by analyzing the envelope spectrum of each modal component,which can effectively identify the locomotive bearing fault type.In order to select the modal components with fault information in each mode and improve the diagnostic efficiency,a feature extraction method based on the kurtosis of square-envelope spectrum for selecting the VMD modal components is proposed,in which the largest component of the square-envelope spectrum of the VMD modal components is the optimal component,the effective information is extracted from the modal components by time-domain analysis,the fault characteristics are extracted by envelope analysis so as to obtain the accurate bearing fault characteristic frequency.