Feature Extraction And Separation Of Bearing Faults Based On Sparse Shift Invariant Theory

With the rapid development of modern industry and science and technology,modern machinery and equipment are becoming more and more large-scale,sophisticated,automated,and intelligent.At the same time,their operating conditions have become more complex and harsh,making them necessary to have high Operational reliability.If some key parts of the mechanical equipment fail,it will not only bring huge economic losses to the enterprise and the country,but may even cause serious safety accidents and casualties.Rolling bearing is one of the key parts of rotating machinery.Because of its special structure and complex working environment,fault damage often occurs.Therefore,in order to ensure the safe operation of mechanical equipment,it is of great significance to study the fault diagnosis of rolling bearing.This paper focuses on the sparse representation theory of signals,conducts an in-depth study on the diagnosing of weak faults and compound faults of rolling bearings,and combines some other algorithms to optimize them.The main research contents are as follows:(1)Knowing the fault mechanism of the research object is the prerequisite for fault diagnosis.Therefore,the structure,fault form and vibration mechanism of rolling bearings are explained,and the characteristics of vibration signals generated when local faults occur in different parts of the bearing are analyzed.After that,the basic theory of signal sparse representation is explained,and several typical sparse coefficient solving algorithms and over-complete dictionary construction methods are introduced in detail.(2)In the rolling bearing service environment,the interference of strong background noise and the signal attenuation caused by the long transmission path make the bearing fault features often appear weak and difficult to extract.For this,the research based on Laplace wavelet and feature sign search algorithm(FSS)Sparse representation method for weak fault characteristics of rolling bearings(LWFSS).At the same time,the modified adaptive empirical wavelet transform(AEWT)is introduced to obtain the morphological parameters of the Laplace wavelet base and preprocess the original signal in order to obtain a better sparse expression effect.Through the analysis of simulation and experimental signals,the results show that the proposed method has a better effect in extracting weak fault features.(3)In order to solve the problem that multiple faults of rolling bearing are coupled and difficult to diagnose,and the resonance frequency bands caused by different faults are overlapped or not,a compound fault diagnosis method of rolling bearing based on LWFSS and multi-point optimal minimum entropy deconvolution adjusted(MOMEDA)is studied.First,use the LWFSS algorithm to analyze the signal.Here,the correlation filtering method is used to obtain the Laplace wavelet basis that best matches the signal.Envelope demodulation analysis is performed on the obtained sparse reconstructed signal.If each reconstructed signal can extract a single fault feature,the separation ends.If a reconstructed signal still contains multiple fault features,MOMEDA is used for processing.In order to verify the effectiveness of the proposed method,a rolling bearing composite fault simulation experiment is designed,and a variety of fault types and conditions are set up,so as to better study the characteristics of bearing composite fault signal.Through the analysis of the composite fault simulation signal and the experimental signal of the bearing,it is verified that the proposed method can effectively separate and extract the composite fault features of the rolling bearing.