Research Of Noise Source Identification Using Bayesian Regularization

Near-field acoustic holography(NAH)is an efficient noise source identification and sound field visualization technology.It can reconstruct the normal velocity of the surface of the sound source and the radiated sound field by measuring the near-field sound pressure of the vibrating body,so it has become a powerful tool for solving noise problems.In this paper,the following researches are investigated on the problems in NAH:To solve the ill-conditioned problem in noise source identification,regularization is used to suppress the influence of errors on the reconstruction effect,the principles and characteristics of common regularization and parameter selection methods are studied.The generalized crossvalidation method is improved,and the calculation time is reduced.The Bayesian method is proposed to decide the optimal regularization parameters.Through the simulation of thin aluminum plate,the different regularization combination methods are compared and analyzed,and the factors affecting the accuracy of reconstruction are discussed in detail.The results show that Tikhonov combined with Bayesian is the optimal one.To solve the problem of identifying large-scale irregular noise sources,a combined NAH method combining Helmholtz equation least squares(HELS)method and inverse boundary element method(IBEM)is proposed,and the specific implementation process of the method is established.Simulation results show that the method can accurately reconstruct the sound field.To solve the problem of identifying the noise source in the mixed sound field,the sound pressure and particle velocity on a single measurement surface are used to derive the sound field separation formula based on BEM.The simulation results of the mixed radiated sound field of a thin aluminum plate show that this method can effectively separate the interference noise and improve the recognition accuracy.Finally,the validity and accuracy of the above theory are verified by experiments.