Research On Fault Characteristic Analysis And Extraction Method Of Rolling Bearing Under Variable Speed Condition

With the development of the times,the working environment of rotating machinery is becoming more and more complex.Rolling bearings,as one of the key components,are facing more and more complex working conditions.Variable speed,as one of the most common working conditions of rolling bearings in actual work,presents new challenges to the fault diagnosis of rolling bearings.The traditional rolling bearing fault diagnosis method under constant speed condition is difficult to accurately identify the fault state under variable speed conditions.Therefore,aiming at the fault diagnosis problem of rolling bearings under variable speed conditions,this study started from the analysis of fault symptoms and aimed to eliminate the difficulties caused by the change of speed for fault feature extraction,and researched fault feature extraction method of rolling bearings under variable speed conditions.The main research work and innovation of this paper are as follows:1)According to the signal characteristics of vibration signals under variable speed conditions,which are different from those under constant speed conditions,a comprehensive analysis is carried out on the changes of fault symptoms in the timefrequency domain of vibration signals under variable speed conditions.Firstly,the vibration characteristics and fault signal characteristics when the local damage fault occurs are analyzed from the bearing structure itself.On this basis,the vibration signal is used as the medium to further analyze the periodic change of the impulse vibration pulse caused by the fault point in the time domain and the fault sensitivity of the time domain index when the speed changes.In the frequency domain,the adaptability of spectrum analysis,envelope spectrum analysis based on fast Fourier transform(FFT)and order spectrum analysis based on order tracking(OT)to vibration signal analysis of variable speed is analyzed and compared.Finally,the characteristic signal changes of timefrequency domain fault diagnosis criteria under constant speed conditions are summarized,and the characteristic indexes suitable for fault identification under variable speed conditions are proposed,and the problems of spectrum analysis and adaptive methods are pointed out.The analysis and verification of experimental data lay a foundation for the subsequent research on fault feature extraction methods.2)Based on the analysis of fault feature symptom changes,aiming at the problems of OT order spectrum,such as excessive noise interference,and fault feature order information is not prominent,a fault feature extraction method based on improved computational order tracking(COT)is proposed.Firstly,to calculate the COT Angle resampling frequency,the determination of the maximum order of the signal depends on manual experience.Considering the relationship between the signal frequency and the order and the significance of the low-pass filter cut-off frequency,the low-pass filter order determination theory is proposed,and the COT Angle resampling theorem is combined to determine the optimal Angle resampling frequency.Secondly,given the problem of noise and harmonic interference in the Angle domain,considering the auto-correlation function(ACF)to eliminate the irrelevant signal,highlighting the nature of the fundamental frequency and the periodicity of the Angle domain signal,based on the Angle domain vibration signal obtained from the optimal Angle resampling frequency,the autocorrelation function(ACF)is introduced to realize the Angle domain noise elimination.After noise removal,the vibration signals in the Angle domain are transformed by FFT,and then the order spectrum is obtained and the fault feature of the order domain is extracted.It is proved by experiments that the constructed ACF-COT model based on filtering order determination theory can extract the prominent fault characteristic order.3)Given the problems that traditional time-frequency analysis methods fail due to the strong amplitude-frequency modulation characteristics and time-variability of vibration signals under variable speed conditions,and it is difficult to extract effective feature information and carry out high-precision classification,an adaptive fault feature extraction and classification method,FVMD-SCN,is proposed under variable speed conditions.Firstly,considering that the variational mode decomposition(VMD)method is difficult to accurately decompose the modal components of time-varying multicomponent signals,based on the existing framework of VMD,a fractional-order Fourier transform(FRFT)was introduced to construct an optimal selection model of modal components and center frequency,namely,FVMD.Secondly,the skewness factor,kurtosis factor,and root mean square value analyzed in Chapter 2 are used to extract prominent fault features under variable speed conditions.Finally,given the low accuracy of adaptive classification of fault features under variable speed conditions,the advantages of random configuration network(SCN)in network structure,learning efficiency,and generalization ability were used to carry out adaptive classification and recognition of fault features.The experimental results show that the proposed method can effectively extract fault characteristics under variable speed conditions and obtain high diagnostic accuracy,which proves the feasibility of the proposed method.