Research On Rolling Bearing Fault Feature Extraction And Diagnosis Method Based On Optimized MCKD And Signal Decomposition Method

Rolling bearing is one of the main core components of the motor.If there is a fault,it will affect the normal operation of the motor in different degrees.It will not only lead to economic losses,but also endanger people’s lives when the problem is serious.Therefore,it is of great significance to carry out in-depth research on the fault diagnosis method of motor rolling bearings,so as to reduce the occurrence of accidents.Aiming at the problem that the traditional signal time-frequency analysis methods lack of adaptive ability and anti-interference ability in processing fault signals,this paper analyzes and compares the signal processing effects of empirical mode decomposition(EMD),empirical wavelet transform(EWT)and variational mode decomposition(VMD).Compared with EMD,EWT and VMD can effectively suppress the mode aliasing in signal decomposition,and have better robustness in signal processing.In order to compensate for the loss of fault characteristics in the transmission path,the signal deconvolution method is introduced.Through the comparison of simulation experiments,the compensation effect of minimum entropy deconvolution(MED)and maximum correlation kurtosis deconvolution(MCKD)on the transmission path of bearing vibration signal is analyzed.In the experiment,it is found that MCKD is more sensitive to the periodic characteristics of the signal,and MCKD will have better effect for the bearing vibration signal.In this paper,the motor bearing is taken as the research object.Aiming at the problem that the fault characteristics of bearing rolling element signal and fault signal under complex working conditions are very weak,and it is difficult to carry out fault diagnosis,this paper combines two signal processing methods of VMD and MCKD,and proposes a weak fault feature extraction method of bearing based on the combination of optimized MCKD and improved VMD.In order to solve the mode number selection problem of VMD,a mode number selection method based on kurtosis mean is proposed.To solve the problem that the processing effect of MCKD is affected by parameters,a parameter optimization method based on power spectrum entropy is proposed.Finally,the processed signal is analyzed by power spectrum to extract its fault characteristics,the optimized method has better performance in dealing with the weak fault of bearing.This paper also proposes a new method of bearing weak fault diagnosis based on improved EWT and MCKD.This method proposes a new spectrum segmentation strategy for EWT,and combines Pearson coefficient and optimized MCKD to extract bearing weak fault features effectively.For the problem that the signal transmission path is long and easy to be interfered due to the sensor acquisition position is not ideal under the actual working condition,and the bearing rolling element fault feature is weak and difficult to extract,a feature extraction method based on rotation frequency separation is proposed,which effectively extracts the bearing rolling element fault feature collected at the base end.