Research On Fault Diagnosis Of Rolling Bearing Based On Full Vector-blind Source Separation

Rolling bearing is the core component of various rotating machinery,the fault of bearing will seriously affect the operational efficiency of mechanical equipment.However,these mechanical equipments usually work in the background of strong noise.The fault vibration fault signals and fault information of rolling bearings are often submerged in noise.The traditional fault diagnosis method of bearing is difficult to extract the characteristic information of the bearing fault from the environment of noise effectively.This paper study the vibration signal of the rolling bearing in the background of noise,combining the advantages of blind source separation and full vector spectrum,and research the effect of blind source separation in the separation of noise and bearing fault signals.Through experiments and simulations,the full vector information fusion is carried out for the fault signal after the separation of signal and noise,and the bearing fault features are extracted to realize the bearing fault diagnosis.The main work is as follows:(1)Introducing the blind source separation of independent component analysis and the theoretical algorithm of inherent time scale decomposition.The method of inherent time scale decomposition is used to carry on multi-scale decomposition for the fault signal to get multiple PRC components,and solve the problem that the number of observation signals is less than the source signal in the process of blind source separation.The flow chart of the blind source separation method of inherent time scale decomposition and independent component analysis is given,and the validity of the separation of signal and noise is indicated by the simulation and experiment.(2)The full vector spectrum analysis based on the homologous double channel signal is introduced by the defective analysis of the fault characteristics of the single channel signal analysis,elaborating the theoretical basis and numerical algorithm.Combine the full vector spectrum and independent component analysis method,a new method of extracting bearing fault feature is proposed for the separation of signals and noise called full vector ITD-ICA.The flow chart of the method is introduced in detail,the experiment indicates that the method of extracting fault feature in bearing is effective.(3)The independent component analysis has some limitations in the processing of nonlinear changing signals.The comparison of simulation between KICA and ICA shows that the nuclear independent component analysis has better nonlinear processing ability than the traditional independent component analysis.A blind separation method of ITD-KICA which used KICA to eliminate information redundancy and extract effective independent information components,the fault signals and noise make to separate.Then the method is combined with the full vector spectrum,the method of full vector ITD-KICA is put forward,and the specific algorithm flow of the method is introduced in detail.The experiment indicates that this method is better to achieve the purpose of reducing the noise,at the same time it can make the bearing fault information more comprehensive and accurate to realize the bearing fault diagnosis.