Research On De-noising Of Train Bearing Wayside Signal Based On Sparse Representation By Doppler Atom Dictionary

The health status of train bearings has a direct impact on the stable,safe and efficient operation for trains.As train bearings,wayside acoustic monitoring is one of the effective means for online monitoring.In this paper,NJ(P)3226X1,a train wheelset bearing in service in China,is taken as the research object.Aiming at the characteristics of wayside signal,the sparse analysis is used as a method to study the de-noising theory of wayside sound signals.In this paper,different kinds of single point failures of train wheelset bearing NJ(P)3226X1 are manually set in service in China,and part of the experimental equipment is introduced,simulate the static and dynamic signal acquisition scheme,and then the field acquisition device is introduced.According to the wayside signal Doppler time-frequency variation,the construction theory and method of Doppler harmonic dictionary and Doppler harmonic-impact composite dictionary are studied,based on the internal time-frequency structure of the wayside signal combining with the matching pursuit algorithm,and according to the different characteristics of spatial distribution for the sound source location,puts forward the optimal atomic filter and the signal reconstruction strategy,realized the wayside sound signal de-noising.Specifically:The first method of this paper is proposed sound signal separation method based on over-complete parameterized doppler modulation harmonic atom dictionary.Specific implementation steps:1.Build the Doppler harmonic atom dictionary;2.The wayside signal is sparse decomposed by matching pursuit algorithm;3.According to the transverse and longitudinal distance from the microphone between the bearing sound source and the resonance frequency band of the sound source signal,select the optimal atom in the projection dictionary,and obtained the de-noising signal by linear combination.The second method of this paper is based on the doppler modulation harmonic-impact composite atom dictionary sparse analysis theory,on the basis of the first method is introduced into doppler modulation impact atoms,build doppler modulation harmonic impact-composite atom dictionary,improve the atoms inner structure of wayside fault signal matching degree,enhancing signal sparse,improve de-noising effect.The validity of the proposed method is verified by simulation and experimental data analysis.Compared with the traditional digital filter,the methods proposed in this paper haven the advantages of eliminating the in-band noise and is expected to be widely used in practical projects.