Fault Feature Extraction Of Deep Groove Ball Bearings Based On Variational Mode Decomposition And Genetic Algorithm-optimized Wavelet Threshold Denoising

Bearings as important mechanical parts in the equipment manufacturing industry affect the operational performance and health status of machinery and equipment.Once a failure occurs,it will not only bring huge economic losses,but may also lead to serious safety production problems and cause major safety accidents.This paper takes deep groove ball bearing as the research object,calculates the fault characteristic frequency of the bearing from two aspects of whether it is subjected to axial force,and uses time domain characteristic parameters,frequency domain analysis and time frequency analysis methods to analyze and study the bearing vibration signal.In deep groove ball bearing fault feature extraction,complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)will have some "spurious" patterns and non-corresponding patterns in the early stage of decomposition,the paper proposes an improved feature extraction method based on improved complete ensemble empirical mode decomposition with adaptive noise(ICEEMDAN)and wavelet packet decomposition(WPT).Validation of the simulated signals shows that the intrinsic mode function(IMF)obtained from the ICEEMDAN decomposition has fewer spurious modes.For the selection of wavelet basis functions,a waveform similarity coefficient analysis method is proposed,and db4 wavelet function with a similarity coefficient λ of 7.4970 and a decomposition layer of 3 is selected.The effectiveness of the method in extracting fault information from the noisy environment is verified through the analysis of the simulated signal of bearing inner ring fault.To address the problem that the ICEEMDAN+WPT feature extraction method has uncertainty in the selection of important parameters,a novel adaptive optimization of an improved feature extraction method variational mode decomposition and genetic algorithm-optimized wavelet threshold denoising(VMD+GA-WT)is proposed.First,VMD decomposition is carried out,and for the difficult problem of VMD parameter selection,a parameter optimization program with small computing amount and better adaptability is proposed.Secondly,the noise is further reduced by using a genetic algorithm to adjust the three parameters in the improved threshold function.The GA-WT method avoids the problems of local oscillations and reduced accuracy of the reconstructed signal due to the defects of the traditional threshold denoising method.Then,the effectiveness of the VMD+GA-WT feature extraction method is verified by simulating bearing inner ring faults,RMSE=0.094 and SNR=6.735.Finally,through the analysis of the measured signal of the deep groove ball bearing with inner ring failure,its spectrum boundary division diagram shows that the VMD decomposition after parameter seeking is more effective,and the characteristic frequency as well as the multiples of the characteristic frequency are accompanied by the speed frequency sidebands around the envelope spectrum diagram after denoising,which indicates that the bearing is in urgent need of replacement,further proving the effectiveness and feasibility of the method to study the denoising characteristics of the deep groove ball bearing.