Research On Vehicle Noise Source Identification Technology Based On Bayesian Compressed Sensing

In the field of transportation,controlling the noise level in the cabin of the vehicle is an important content to improve passenger comfort.The key to controlling noise is to accurately identify the source of the noise.At present,the method of identifying the noise source through sound field reconstruction technology is more mature.Among them,the near-field acoustic holography technology uses the holographic data measured near the surface of the sound source to reconstruct the sound field.It contains more "evanescent wave" components of sound field detail information,which makes it possible to break through the limitation of Rayleigh wavelength on reconstruction resolution.Therefore,near-field acoustic holography technology is a powerful sound field visualization tool,and it is becoming more and more mature in the application of noise source identification.However,to obtain more accurate high spatial resolution reconstruction results through near-field acoustic holography,it is necessary to satisfy the Nyquist sampling theorem to arrange a large number of sampling points on the holographic surface,which greatly increases the cost and difficulty of the experiment,and also reduces the cost.The computational efficiency of the algorithm severely hinders the promotion and application of near-field acoustic holography in practical engineering.Therefore,to overcome this shortcoming of the technology,it is necessary to study the problem in theory and propose a new algorithm.This can not only promote the development of near-field acoustic holography,but also greatly promote the use of near-field acoustic holography.Application and promotion in the field of noise source identification.In view of the above problems,the main research contents of this article are as follows:First,there is the problem of ill-posedness of the source strength solution in the sound field reconstruction process.A new solution is proposed for the regularization method used in the solution-Bayesian regularization.Moreover,the comparison and analysis between simulation and traditional regularization methods are performed to illustrate the feasibility of the Bayesian sound field reconstruction method and its existing advantages.Secondly,in order to solve the problem that the spatial resolution and maximum analysis frequency of near-field acoustic holography are limited by the sampling theorem,a sound field reconstruction method based on Bayesian compressed sensing theory is proposed.The singular value decomposition technique is used to obtain the basis function,and the sparseness of the sound pressure signal under the basis function is analyzed using the signal sparse representation theory,and the sound field reconstruction problem is modeled.The effectiveness and noise immunity of the method were analyzed by simulation and the method was verified with experiments.Finally,according to the rigid constraint condition of the sound source surface,the interference sound in the non-free sound field and the scattered sound on the surface of the sound source are quickly removed.Combined with the Bayesian compressed sensing theory,a new model of free sound field reduction is established to provide the necessary pre-processing technology for implementing near-field acoustic holography in practical engineering applications.