Research On Wayside Acoustic Fault Diagnosis Of Train Bearing Based On Rectangular Array Optimal Spatial Filter

Railway transportation plays an important role in the national economy.Wheelset bearings are the key parts of trains and it is of great significance to carry out on-line condition monitoring research on them.The wayside acoustic detection technology uses microphone arrays installed on both sides of the rail to collect the sound signals emitted by the wheelset bearings when the train is running and realizes condition monitoring and fault diagnosis through signal processing.It has the capability of non-contact measurement,low cost and early warning.However,the characteristic of heavy noise of the wayside signal seriously affect the accuracy of the diagnosis.To solve this problem,this paper introduces a rectangular array and proposed a spatial filtering algorithm based on Minimum Variance Distortion-Less Response(MVDR).Directional denoising is realized by designing an Optimal Spatial Filter(OSF).The algorithm is combined with sparse filtering algorithm to further denoise from the perspective of signal components,which provides a new idea for the problem of strong noise of wayside signal.The specific work is as follows:An MVDR algorithm based on rectangular array is designed to realize directional de-noising of wayside signal.Through simulation and experimental comparative analysis,the effectiveness of the algorithm and its advantages over the existing linear array technology are verified.First,the limitations of the existing single microphone and linear array solutions are analyzed.Secondly,a 3×5 rectangular microphone array is designed,and a directional filter is designed based on the MVDR algorithm to achieve directional denoising.At the same time,based on instantaneous energy analysis and resampling technology,Doppler distortion correction is achieved.Finally,simulation and experimental comparative analysis verify the superiority of the proposed method over the existing linear array method.The aforementioned method based on rectangular array spatial filtering achieves a good result,but due to the imperfection of the filter,there is still some noise in the signal after spatial filtering,and it is difficult for the band-pass filter to eliminate the in-band noise.A denoising strategy combining array spatial filtering and sparse filtering is proposed to achieve further denoising from the perspective of signal components.First,the basic principles of sparse decomposition are elaborated in detail,and an over-complete compound dictionary of Doppler modulation is constructed.Secondly,a sparse decomposition algorithm is designed based on the compound dictionary and the matching pursuit algorithm,and the sparse filtering of the target sound source signal is achieved through atomic parameter selection and optimal atoms reconstruction.Then,a spatial filter is constructed to filter the sparse filtered signal.Finally,the effectiveness of the proposed method is verified by simulation and experimental signals.The results show that combining spatial filtering with sparse filtering can achieve better denoising.