Research On Vibration Fault Diagnosis Method Based On Singular Spectrum Decomposition

With the progress of society and the rapid development of science and technology,modern production is more and more inseparable from a variety of rotating machinery.Gears and bearings,as key components of rotating machinery,directly determine the normal operation of the entire mechanical system,and their safety has a great impact on the normal operation of the entire equipment.Deeply developing the research of gear and bearing fault diagnosis and condition detection technology has important practical significance for ensuring the safe and stable operation of equipment and avoiding the occurrence of major safety accidents in production process.Therefore,this paper takes gears and bearings as the research object,combines singular spectrum decomposition with Wigner-Ville distribution,1.5-dimensional spectrum,multi-scale dispersion entropy and support vector machine,and uses a variety of algorithms to realize fault diagnosis of gears and bearings.The work and achievements of this paper are as follows:The performance of singular spectral decomposition(SSD) and Wigner-Ville distribution(WVD) in time-frequency analysis is studied.A new adaptive fault diagnosis method based on singular spectral decomposition(SSD-WVD) is explored and applied to gear fault diagnosis.Firstly,the original vibration signal is decomposed into several SSC components by using the good adaptive decomposition ability of singular spectrum decomposition.Then,the kurtosis value of each SSC component is calculated,and the optimal SSC component is selected according to the kurtosis value.Finally,the Wigner-Ville distribution of the optimal SSC component is calculated.Through the analysis of measured signals,it is found that SSD-WVD method has a good diagnostic effect on gear fault.In order to solve the problem that the optimal number of decompositions and the selection of the optimal components can not be achieved when the singular spectrum decomposition algorithm decomposes the signals,the feature energy ratio(FER) is used to improve the SSD method.A bearing fault diagnosis method based on the combination of singular spectrum decomposition and 1.5-dimensional spectrum is adopted to solve the problem that the fault feature information of rolling bearings is difficult to extract by applying the good decomposition characteristics of SSD.Experiments show that the proposed method is effective in bearing fault feature extraction.Faced with the problem of bearing condition recognition in complex working conditions,a bearing fault recognition and classification method based on singular spectral decomposition and multi-scale dispersive entropy(MDE) combined with support vector machine(SVM) is studied.First,the original signal is de composed by singular spectral decomposition algorithm,and the components with abundant fault feature information are selected as the principal component of analysis,and then the principal component is decomposed by MDE algorithm.After multi-scale feature extraction,feature vectors are constructed.At last,feature vectors are input into SVM classifier.After training and testing,bearing state is recognized by pattern recognition,and bearing fault diagnosis and classification are realized...