Research On The Reconstruction And Separation Of The Sound Filed Radiated By The Sources With Arbitrary Shape Based On Sparse Sampling

Nearfield acoustic holography(NAH)is an advanced tool for identifying noise sources and visualizing sound filed.With the measurement of the sound field close to the sound sources,the acoustic information on the source surface can be reconstructed and that in the three-dimensional sound field can be predicted by using the NAH technique.The technique requires the interval between two adjacent spatial sampling points satisfying the Nyquist criterion.Otherwise,the reconstruction accuracy of NAH will decrease due to the spatial aliasing.However,the decrease of spatial sampling interval means that the number of spatial sampling points is increased,which will lead to a high hardware cost if using the snap-shot method with a large array or heavy workload of measurement if using the scanning measurement method.In this dissertation,the theory of the signal sparse representation and sparse regularization was applied to NAH to deal with the problem of measurement efficiency of NAH.The acoustic radiation modes of the source surface velocities were constructed and the sparse regularization was applied to the inverse boundary element method(IBEM).Considering the computation efficiency of IBEM,the sparse regularization was applied to the equivalent source method(ESM)and the performance of the ESM based on the sparse sampling(S-ESM)was analyzed.Then the reweighted iterative algorithm was applied to the S-ESM to improve the spatial resolution of the equivalent source strengths at low frequencies and low SNRs.For the case that the S-ESM was not suitable for the spatially extended sources,the acoustic radiation modes of the equivalent source strengths were constructed and the ESM based on the sparse sampling and the acoustic radiation modes of the equivalent source strengths(SM-ESM)was proposed.Meanwhile,based on the bases used in the S-ESM and SM-ESM,a redundant dictionary of the equivalent source strengths was constructed and the ESM based on sparse sampling and the redundant dictionary(SD-ESM)was proposed.The method could provide high reconstruction accuracy whenever the source was spatially sparse or spatially extended.Besides,the sound field separation technique based on sparse sampling and ESM was proposed to account for the effect of the disturbing sources on the other side of the hologram surface.The method could provide relatively high separation accuracy with reduced sampling points.The dissertation was organized as follows:In chapter 1,the background and the development of NAH were reviewed,and the problems related to the spatial resolution,measurement efficiency and cost were discussed.Then the application of the theory of the signal sparse representation and sparse regularization in NAH was introduced and the existing problems in these applications were presented.Finally,the main research objectives were determined.In chapter 2,the acoustic radiation modes of the source surface velocities were constructed and were used as the sparse basis of the source surface velocities.Then the sparse regularization was applied to the IBEM and the IBEM based on sparse sampling(S-IBEM)was proposed.The results of numerical simulations and experiment demonstrated that the proposed method could provide high reconstruction accuracy of the source surface velocities with reduced sampling points.In chapter 3,the performance of S-ESM was firstly analyzed.Then the reweighted iterative algorithm was applied to the S-ESM and the ESM based on sparse sampling and reweighted iteration(SI-ESM)was proposed.Moreover,the SI-ESM was further extended to the identification of the noise sources in the three-dimensional space.Finally,the effectiveness of the proposed S-IESM was demonstrated through numerical simulations and experiment.In chapter 4,for the case that the S-ESM was only suitable for the spatially sparse sources,the acoustic radiation modes of equivalent source strengths were constructed and the SM-ESM was proposed to solve the problem of the sound field reconstruction of the spatially extexeded sources.Meanwhile,the sparse bases of S-ESM and SM-ESM are combined to form a redundant dictionary of the equivalent source strengths and the SD-ESM was proposed.Finally,the effectiveness of the proposed methods was analyzed and demonstrated through numerical simulations and experiment.The results indicated that the SM-ESM could provide the reconstruction accuracy of the sound field radiated by the spatially extended sources with reduced sampling points and SD-ESM was more flexible for different types of sources.In chapter 5,the signal sparse representation and sparse regularization was applied to the ESM-based sound field separation technique to further account for the effect of the disturbing sources on the other side of the measurement plane.Then the S-ESM-based,SM-ESM-based and SD-ESM-based sound field separation techniques were proposed according to the selection of the sparse basis of the equivalent source strengths.The effectiveness and performance of the proposed methods were analyzed and demonstrated through numerical simulations and experiment.In chapter 6,summaries to the dissertation were drawn and the topics needing further research were proposed.