Research On Fault Diagnosis And Trend Prediction Method Of Rolling Bearing

With the rapid development of industrial technology,The rotating machinery is developing towards the direction of large-scale,high-speed,complicated and intelligent.As a key universal component widely used in rotary machinery—rolling bearings,whether its running state is normal,is seriously related to the running state of the whole equipment.In practical applications,as the running condition of the rolling bearing is complex and the working environment is overloaded with high speed,the rolling bearing is one of the most easily damaged mechanical parts.The rolling bearing is taken as the research object in this thesis.In this paper,a number of new methods in modern signal processing are applied to three key problems: fault feature extraction of rolling bearing,compound fault diagnosis of rolling bearing,rolling bearing running status monitoring and performance degradation trend prediction.The main works are as follows:(1)Aiming at the problem that rolling bearing fault characteristics are difficult to extract,a fault feature extraction method based on IVMD and energy operator demodulation of three point symmetrical differencing is proposed.On the basis of introducing the basic concepts and principles of VMD,the signal decomposition ability of VMD is discussed through the simulation signal.Facing on the problem of setting the VMD parameter K,an improved variational mode decomposition(IVMD)method is proposed,which takes the minimum value of the spectral correlation coefficient between the BLIMFs and the original signal as the decomposition stop condition,to realize the adaptive determination of decomposition level K in VMD decomposition process.Through simulation and experimental signal analysis,it is shown that the combination of IVMD and energy operator demodulation of three point symmetrical differencing can extract the fault characteristic frequency of rolling bearing very well.(2)Focused on the problem that the compound fault of rolling bearing is difficult to diagnose accurately,a new method of compound fault diagnosis for rolling bearing based on MOMEDA is proposed.Firstly,the basic principles of various time domain deconvolution methods are introduced in this paper.Then,the simulation signal is used to discuss the ability of the MOMEDA method to extract the continuous periodic shocks in the signal.On this basis,the multi-point kurtosis is used as the measure of the fault feature,and the MOMEDA is used to diagnose the compound fault of the rolling bearing.(3)In view of the problem that rolling bearing performance degradation trend is difficult to predict effectively,a prediction method of rolling bearing performance degradation trend based on VMD energy entropy fuzzy granulation and SVM is proposed.This method takes VMD energy entropy as a characteristic index of rolling bearing performance degradation,and combines fuzzy information granulation and SVM to achieve accurate prediction of rolling bearing performance degradation trend.In this paper,the influence of the parameter K of the VMD method on VMD energy entropy is also analyzed.The validity and accuracy of the method mentioned in this paper is proved by the bearing full life experiment data..