Research On Passive Detecting Technology Using Acoustic Vector Sensor Array For Underwater Small Carrier

With the development of unmanned underwater vehicle?UUV?demanding of high speed and large voyage,the passive sonar system installed on the UUV is not only restricted by the aperture of the platform,the level of self-noise and background interference received by the sonar array is also enhanced.Therefore,the strong noise background,the near-field interference,the array error and lack of snapshot number all put forward higher request on array signal processing.In order to meet the high resolution requirement for the direction estimation of far field targets in the non-ideal complex environment,this paper takes the advantages of good performance of the acoustic vector array,such as noise suppression and array aperture expansion.Combined with the practical environment of underwater acoustic equipment and the requirement for array signal processing in non-ideal environment such as high level of noise and interference,theoretical and experimental studies on passive direction finding and array calibration technologies are carried out in this paper.The detailed contents of this paper are as follows:1.In order to solve the high resolution direction finding problem of the far field target under low signal to noise ratio using acoustic vector array under the limited aperture,a sparse reconstruction azimuth estimation algorithm based on the optimal weighted signal subspace fitting is proposed.When the classical sparse reconstruction direction of arrival estimation algorithm L₁-SVD is applied to the acoustic vector array,it is found that L₁-SVD is equivalent to a weighted subspace fitting problem with space sparsity constraint,but not weighted by the optimal weight matrix.After reviewing the problem of optimal weight matrix with the minimum mean root mean square error in the theory of weighted subspace fitting,the weighted subspace fitting problem with optimal weight is converted to a sparse reconstruction problem of L1 norm constraint.The algorithm constructs the noise constraint coefficient by estimating the statistical properties of the residual.Simulation results and experimental data on lake show that the algorithm is insensitive to the prior information of the number of sources,and it can effectively improve the performance of DOA estimation under low signal-to-noise ratio.2.In the course of navigation,the UUV platform will produce near field noise source interference.In view of the problem of far field target azimuth estimation using acoustic vector array under the interference of near field source,a separation technology for mixed far-field and near-field sources is proposed.On the basis of the analysis of the correlation difference between the steer vector of the far field plane wave and that of the near-field spherical wave,the boundary of the far and near field is bounded.A sparse representation model of far field and near field is constructed.Sparse Bayesian model is introduced to fit the array output data and constrain the spatial sparsity.Taking advantage of the high resolution characteristics of sparse reconstruction algorithm,the energy of near-field interference is constrained in the near-field steering vector dictionary,so as to ensure the high resolution estimation performance of far-field targets.The algorithm does not require the number of sources and noise power as a priori information input,thus avoiding the adjustment of parameters.Simulation and experimental data on lake verify the superiority of the algorithm in the presence of near field strong point source interference.3.Most of the high resolution azimuth estimation algorithms assume that the noise field is an isotropic Gaussian background.However,the noise sources on UUV are widely distributed and under the effect of near-field volume noise sources,the received noise spectrum of the acoustic vector array has a certain spatial directivity,which dose not meet the assumption of isotropy.In view of the problem of azimuth estimation in the presence of spatial nonuniform noise,the excellent properties of prolate spheroidal wave function in the finite bandwidth signal fitting are extended to the space and a linear noise model which is more suitable for describing the spatial nonuniform noise is proposed.The reconstruction of spatial power at all angles and the noise parameters is solved by an iterative approach under the framework of sparse Bayesian learning.The joint iterative solution improves the accuracy of direction of arrival estimation of far-field sources in the spatial directional noise field.The simulation results verify the effectiveness of the algorithm,and the performances of estimated accuracy and resolution for different algorithms in the non isotropic noise environment are compared and analyzed.4.The high resolution azimuth estimation algorithm based on acoustic vector array generally requires that the array manifold is accurately known,but various uncertainties in practical environment may lead to the existence of array error.On the basis of the analysis of the mismatch degree between real steer vector and ideal steer vector caused by array shape error and attitude error,along with their influence on the spatial spectrum,the active correction algorithm for the position error of the vector array element is put forward,which is easy to obtain by using the position error of the array element.A sparse signal reconstruction perspective for vector array shape calibration is put forward.Using the characteristics of the upper and lower bounds of the array element position error,the overcomplete array output model with array element position is obtained by discrete sampling of acoustic range difference between different array elements.Thus,the problem of the position estimation of the array element is converted to a sparse reconstruction problem.The algorithm is not subject to the assumption of small position error and can be used for the array with arbitrary formation.Simulation results verify the effectiveness of array element position error estimation,and prove that the calibration process improves the performance of high resolution direction of arrival estimation algorithms.