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

Rolling bearings as the important components of mechanical rotating machinery,its running state fault or not working directly affects the safe and stable operation of the whole mechanical systems.To meet different needs of production demands,the rotating speed of rolling bearings has both constant speed and variable speed in the practical operation.Therefore,Extracting effective fault characteristics from fault signals running different status is very important for the fault diagnosis of rolling bearings.The key to rolling bearing fault diagnosis is extract the fault information of vibration signals,so selecting appropriate signal decomposition method is necessary.At present,there are many representative signal decomposition methods,for instance,empirical mode decomposition(EMD),ensemble empirical mode decomposition(EEMD),local mean decomposition(LMD),and its a series of improved algorithms.However,All these methods still have many hard to resolve disadvantages,such as mode mixing problem and end effect.So,It is particularly important to adopt a new signal decomposition method in the field of bearing fault diagnosis.Variational Modal Decomposition(VMD)is a new signal processing method creatively proposed by Konstantin Dragomiretskiy in 2014.This method has a sound theoretical basis and mathematical framework.In this thesis,Based on the variational modal decomposition algorithm research and applying it in vibration signals analysis of rolling bearings.The diagnosis methods of different running status rolling bearing faults are studied.The main contents are as follows:Firstly,the first chapter describes the research background and significance briefly,the fault diagnosis method,bearing structure,vibration mechanism and calculation method of fault characteristic frequency are comprehensively elaborated.On the base,it puts out the main study methods.By analyzing the simulation signal with VMD and EMD,it is proved that the method of VMD is superior to the EMD method in anti-modal aliasing and noise robustness.the influence of different VMD parameter Settings on signal decomposition is explored through simulation signal,a reasona-ble parameter setting scheme is proposed based on the correlation coefficient and center frequency of each component.Secondly,Aiming at the problem that the early vibration signal fault information weak and difficult to extract,a fault diagnosis method based on cross-correlation function and VMD is proposed.the early fault vibration signal by cross-correlation analysis,which can eliminate the background noise and highlight the fault impact component.Then the VMD analysis is performed on the noise reduction signal,the components with higher kurtosis and mutual information values are demodulated.By analyzing simulated vibration signal of rolling bearing with out-race fault and measured vibration signal of rolling bearing with inner-race fault or rolling ball fault,The results show that the proposed method can effectively clear the noises and extract the fault feature information.Thirdly,the rotating speed of rolling bearing in actual rotation is variational,because of the on-off or load change.In order to realize the rolling bearing fault diagnosis under variable speed conditions,incorporating the merits of order spectrum kurtosis algorithm and VMD,which is used to extract the fault characteristics of rolling bearing with variable speed.compared with order envelope and order-VMD analysis,The experimental results show that the proposed method is efficiency and veracity in fault feature extraction under variable speed conditions.