Mechanical Fault Diagnosis Method Based On Null Space Pursuit Algorithm

This dissertation was supported by the National Natural Science Foundation of China(No.51675258,51265039,51075372),the State Key Laboratory of Mechanical Transmissions of Chongqing University(SKLMT-KFKT-201514),the National key R & D program(No.2016YFF0203000),the Key Laboratory of Digital Signal and Image Processing of Guangdong Province(2014GDDSIPL-01)and the Science and Technology Project of Jiangxi Provincial Education Department(No.GJJ12405,GJJ150699).In this paper,Null Space Pursuit method is applied to the mechanical fault diagnosis,and a mechanical fault diagnosis method based on Null Space Pursuit algorithm is proposed,and some innovative achievements are obtained.The main results are as follows :1.The proposition of the subject and its research significance is discussed,the research status of null space pursuit algorithm at home and abroad and the research status of the pursuit algorithm in fault diagnosis are reviewed.On this basis,the research content and innovation of this thesis are given2.The theoretical knowledge of null space pursuit(NSP)algorithm is discussed,including local narrow band signal,linear differential operator and integral operator.The difference between these two operators is compared,and the local narrow band signal under different operators is obtained.The null space pursuit algorithm based on parametric differential operator is analyzed in detail,and the TK energy operator with the same narrow band signal extraction is also compared theoretically.Finally,the null space pursuit algorithm of integral operator and differential operator is compared and analyzed by simulation experiment.The simulation results show that these two kinds of operators can extract the local narrowband signal well.The theory of null space pursuit algorithm studied in this chapter is the theoretical basis of the whole thesis.3.The method of mechanical fault diagnosis based on null space pursuit algorithm is proposed and the diagnosis process is given and compared with EMD.The simulation results show that the proposed method can effectively decompose the multi-component non-stationary signal and restore the spectral characteristics of the original signal,and do not produce too much mode.Null space pursuit algorithm is superior to the EMD method,the method of noise suppression is also very strong.Finally,the proposed method is used to analyze the inner and outer ring faults of the rolling bearing.The experimental results can effectively detect the characteristic frequency of the inner and outer ring faults,and further verify the effectiveness of the proposed method.4.The definition and algorithm of Generalized S Transform(GST)is discussed.The Generalized S Transform is a non-destructive time-frequency analysis method,which can adjust the wavelet basis by controlling the parameters so as to obtain better flexibility and higher time-frequency resolution.Null space pursuit algorithm can use the local singular linear operator to adaptively extract the decomposition algorithm of narrow band signal.In this paper,combined with the advantages of NSP and GST,a hybrid signal decomposition method based on NSP and GST is proposed,which is called NSP-GST algorithm.The theory of NSP-GST algorithm is discussed in detail,and compared with the mixed signal decomposition method(NSP-ST)based on NSP and standard S transform(ST).The simulation results show that the proposed method is superior to the NSP-ST algorithm and successfully achieves the separation of the frequency domain of the simulation signal and obtains better time-frequency resolution.Finally,the NSP-GST method is used to diagnose the rotor rubbing fault.The experimental results show that the proposed method is effective and can detect the characteristics of rubbing fault.5.The theory,algorithm and the existing problems of EMD are discussed.The EMD method is modified and the convergence is proved for the defects of EMD method.At the same time,the reason for the limitations of NSP is analyzed theoretically.On the basis of the limitation of EMD and NSP,a fault diagnosis method based on EMD-NSP is proposed,compared with the traditional EMD and NSP decomposition methods.Simulation results show that the proposed algorithm is effective and the EMD-NSP method is superior to EMD and NSP decomposition method.Finally,the algorithm is used to diagnose the fault of the bearing,which can effectively extract the fault characteristic frequency of the rolling bearing fault,and further verify the effectiveness of the method.