Rolling Bearing Fault Diagnosis And Remain Useful Life Prediction Research Based On Vibration Signal Analysis

Rolling bearing is an important component of rotating machinery,it's well working is an guarantee for the safe operation of the mechanical system.It's important to do research on rolling bearings fault diagnosis and remain useful life prediction.On the basis of knowning the research background and significant,this paper intruduce the adaptive variational mode decomposition and parameter estimation method to realiz rolling bearing.fault diagnose and remain useful life prediction.The main contents of this paper are as follow:Firstly,insight into theThe development and current status of rolling bearing fault diagnosis,analysis the basic fault diagnosis process and the common method.And variational mode decomposition(VMD)method is used for the adaptive decomposition of bearing vibration signals,get the signal components that contain the main fault information.Contrapose the point that decomposition number of intrinsic mode function(IMF)components in the method will affect the decomposition result,researched a method which based on fast spectral kurtstom to set the number of IMF.The mothed apply spectral kurtosis to estimate the bandwidth of the optimal frequency band,astrict the proximity of the center frequency of IMF while its' quantity.Increased and btain the reasonable value of insrinsic mode function.Secondly,apply VMD to decompose the bearing signal and compare the result with empirical mode decomposition.The sensitive IMFs are selected according to the kurtosis-correlation index.Extract the time domain and frequency domain features of sensitive IMF.The classification of extract data is realized by random forest classifier.Finaly,in order to predict the remain useful life of the rolling bearing,by analyzing the life cycle and vibration signal of the bearing,use root mean square as health index to describe the change of bearing status during its life cycle.Point on the problem that double exponentia model have too much parameters and the initial value influence the estimate result,research the method that decrease the number of estimate parameters.Establish the bearing's empirical exponentia recession model.Particle filter is used to estimate the parameters of recession model and predict the future state of the bearing according to the estimated results.