Research On Fast And High-resolution Bearing For Towed Line Array Based On Sparse Bayesian Learning

Towed line array sonar has the advantages of a large aperture and being far away from the platform radiation noise interference,and works at low frequency for long-distance detection,which is widely used in ocean exploration and ocean resource development.However,there is a problem that the bending of towed line array during the maneuvering turn of the platform leads to the degradation of bearing performance.Therefore,this paper focuses on the high-resolution direction of arrival(DOA)estimation based on sparse Bayesian learning(SBL),combined with array structure optimization to achieve algorithm acceleration.In this paper,we first perform parabolic modeling of the distorted array shape and carry out the research of the SBL bearing algorithm under the sparse reconstruction theory in combination with the traditional DOA estimation algorithm.To address the problem of high computational complexity of the SBL algorithm,the real-value sparse Bayesian learning(RSBL)algorithm is proposed to convert the signal model in the complex domain to the real domain and reduce the computational load of iteration.To address the problem of slow convergence of the SBL algorithm,the fast convergence sparse Bayesian learning(FCSBL)algorithm is proposed to approximate the hyperparameter controlling the prior variance of signal and reduce the number of iterations required for convergence.In addition,considering the real-time change of array shape of towed line array during platform maneuvering,the adaptive bow FCSBL(ABFCSBL)algorithm is proposed to quickly and jointly estimate the array shape and target orientation.The large aperture towed line array has many elements,and the high matrix dimension in the iterative process of the FCSBL algorithm leads to the increase in computational load.To this end,the idea of sparse array is extended from linear array to curved towed line array,and coprime array and nested array are used instead of dense uniform array to further reduce the computational complexity of the ABFCSBL algorithm.The number of elements is reduced and the computational efficiency of the bearing algorithm is improved on the premise that the estimation accuracy of array shape and DOA is similar to that of dense uniform array.Numerical simulations verify the high-precision DOA performance of the RSBL algorithm and the FCSBL algorithm,and evaluate the target localization capability of the ABFCSBL algorithm during platform maneuvering.The experimental data of the South China Sea and the west coast of Italy verifies the effectiveness and practicality of the above algorithms.