Study On Fault Diagnosis Of Rolling Element Bearings Based On Resonance-based Sparse Decomposition

Rolling bearing plays a pivotal role in mechanical system,and its stability is a necessary factor for the continuous operation of the whole system.Therefore,it is of great significance to analyze the faults of rolling bearing.The main content of fault diagnosis lies in the extraction of fault information and pattern recognition.Fault information is usually expressed as state feature parameters,while pattern recognition is essentially a process of identifying and classifying different types of faults.The thesis takes rolling bearings as the research object.A fault diagnosis method for rolling bearings based on Resonance-based Signal Sparsity Decomposition(RSSD)is proposed,focusing on the application of parameter-optimized RSSD in fault information extraction,and the application of RSSD and permutation entropy in identifying fault types of rolling bearings.Firstly,a noise reduction method based on Tunable Q-factor Wavelet Transform(TQWT)and improved wavelet threshold is proposed.Aiming at the problem that it is difficult to directly extract the fault features from the background noise,the correlation kurtosis is used to realize the adaptive selection of TQWT parameters and the selection of sub-bands,and de-noised the selected sub-bands though an improved wavelet threshold function,and then reconstruct the sub-bands to extract the faulty information.Using the simulated signal,it is proved that the method can more obviously separate the fault information hidden in the signal.Secondly,a fault information extraction method based on parameter-optimized RSSD is proposed.Aiming at the problem of poor matching between artificially selected parameters and signals,an improved Particle Swarm Optimization algorithm is applied,and the ratio of the kurtosis of the low resonance component to the correlation coefficient of the resonance component is used as the fitness function,then the RSSD parameters are optimized to make more fault information is separated into the low resonance component.Through the processing of the bearing fault test signal,it is proved that the method can effectively separate the fault impact in the vibration signal.Finally,a method for identifying bearing fault types based on RSSD and multiscale permutation entropy is proposed.Aiming at the problems of asymmetry of sample points and missing information caused by the coarse-graining process of multi-scale permutation entropy,the coarse-graining process is improved to obtain generalized compound multi-scale permutation entropy.First,perform parameter-optimized RSSD on the signal,and then use the generalized composite multi-scale permutation entropy of low resonance components as samples,and use the sample data to train the support vector machine to predict and classify different samples.This method is applied to bearing experimental signals to verify that the method has higher accuracy in identifying the type of rolling bearing faults.It can be concluded from the experimental results that the method proposed in this paper can effectively identify and classify bearing faults and provide a new idea for the fault diagnosis of rolling bearings.