Rolling Bearing Fault Diagnosis Based On Chaotic Singular Spectrum And Support Vector Machine

Rolling bearings is widely used mechanical parts in rotating equipment,but easily damaged part.Its operating conditions directly related to the safety of the entire equipment and functional realization.The life of bearing is larger discrete due to its production processes and the working environment.Therefore regular maintenance and replacement will not only cause the waste of resources,but also influence production efficiency.In this paper,the method of rolling bearing fault diagnosis is based on the technique of pattern recognition,grounding the method of chaotic singular spectrum and support vector machine by rolling bearing vibration signal.The main work of the paper are as follows:(1)First,the basic structure,fault form and the vibration principle of the rolling bearing are expounded,as well as the commonly method bearing fault diagnosis.And the research method and work content are established in this paper.(2)Second,the reasons of chaos formation and the characteristics of chaos attractors are introduced.The theory of phase space reconstruction,the common algorithm of reconstructing parameter selection and several kinds of chaos statistics are expounded.These theories and methods are critical for the qualitative and quantitative analysis of chaotic system,which provides a powerful tool to understand chaos and analyze chaotic time series.And the chaotic characteristics of the bearing vibration signal are quantitatively analyzed with the maximum Lyapunov exponent.(3)Then,the method of selecting the wavelet threshold based on the wide spectrum characteristic of chaos is proposed based on the projection of the space of the wave filter bank.By carrying out simulation and numerical verification for the time series of typical Lorenz and Duffing systems in chaotic system,the results show that this method has better noise reduction effect than the other wavelet threshold selection method and can separate the useful signal from the original signal effectively.The power spectrum of the inner and outer fault diagnosis signals is analyzed,which shows that the new threshold method can effectively separate the useful signals from the original chaotic signals.(4)Next,the definition of chaotic singular spectrum and the basic theory of support vector machine are given.The stability and anti-noise ability of chaotic singular spectrum are analyzed by the typical chaotic system.Then some characteristics of chaos singular spectrum,that the singular spectrum of the characteristic space and the noise platform is explained by the perspective of functional analysis and the description of the spatial geometry based on the variance maximization is described by the geometric spatial perspective,are explained theoretically.It shows that the chaos singular spectrum is a quantitative description of the spatial structure of the chaotic attractor,and it has a certain anti-noise interference ability.Therefore,it uses the chaotic singular spectrum as the characteristic quantity of chaotic system and carry out the characteristic analysis for the vibration signal of different bearing parts,which shows that the chaotic singular spectrum can be used as the effective characteristic of bearing vibration signal in this paper.(5)Finally,the fault data of rolling bearing of the Western Reserve University is taken as the object of study.The experiments are carried out by the different damage degree and different fault parts of rolling bearing.The results show that the new wavelet threshold for noise reduction,chaos singular spectrum as characteristic eigenvector and support vector machine as pattern recognition are able to achieve good fault diagnosis effect.(4)the fault data of rolling bearing of the Western Reserve University is taken as the object of study.The experiments are carried out by the different damage degree and different fault parts of rolling bearing.The results show that the new wavelet threshold for noise reduction,chaos singular spectrum as characteristic eigenvector and support vector machine as pattern recognition are able to achieve good fault diagnosis effect.