| Underwater acoustic array signal processing has always been a core research topic in underwater acoustic signal processing and plays a crucial role in marine environment detection,underwater target positioning,and underwater communication.The deployment of large-scale towed line array sonars,i.e.,towed line arrays,has led to a significant increase in power consumption,system complexity,and cost.In response to these challenges,the sparse array optimization design method was proposed,with the primary goal of optimizing the position of array elements to reduce the highest sidelobe level(Peak Side Lobe Level,PSLL)while reducing system complexity and manufacturing costs.This paper aims to study and improve the optimization design algorithm of sparse arrays and explore further the direction of arrival(DOA)estimation algorithm based on sparse arrays,in order to fully exploit their unique advantages.The random search algorithm is widely used in sparse array optimization design to address global optimization performance degradation in large-scale array optimization.This paper proposes an improved random search algorithm that addresses multiple constraints in the optimization of sparse linear arrays.Specifically,the PSLL of the sparse sensor array is used as the fitness function,and the algorithm is applied to the sparse linear array optimization model with constraints.The proposed algorithm improves the quality of the initial position by introducing a hybrid chaotic map to prevent the algorithm from falling into a local solution.Additionally,it implements various strategies to improve the mathematical model of the algorithm and reduce the probability of the algorithm falling into a local optimum,thereby improving the optimization speed and convergence accuracy.Finally,the proposed algorithm improves boundary constraints to enhance global search ability.The paper verifies the superiority and feasibility of the improved random search algorithm under multiple constraints through extensive simulations,demonstrating its effectiveness in the optimal design of sparse arrays.In underwater target azimuth estimation,factors such as non-ideal receiving channels and oceanic noise influence the received noise,often resulting in colored noise.However,when the array elements are sparsely arranged,the colored noise can be simplified to non-uniform noise.To address the issue of decreased DOA estimation performance under non-uniform noise,this thesis proposes an improved robust SBL off-grid DOA estimation algorithm based on the theory of Sparse Bayesian Learning(SBL).The algorithm utilizes an improved iterative method to estimate the nonuniform noise covariance matrix,thereby reducing the impact of non-uniform noise.Additionally,the dynamic grid method is employed to dynamically update the sampling grid to solve the off-grid error in the model.Finally,a step-by-step update strategy is adopted to enhance the estimation accuracy of dynamic grid updates and reduce the calculation load of the algorithm.Through extensive simulations,this paper demonstrates that the improved robust SBL off-grid DOA estimation algorithm achieves robustness and high resolution in DOA estimation.This thesis validates the proposed improved random search algorithm under multiple constraints and the improved robust SBL off-grid DOA estimation algorithm through South China Sea experimental data.The improved random search algorithm is applied to the sparse optimization design of large-scale towed line arrays,achieving the lowest PSLL compared to other random search algorithms.The improved robust SBL off-grid DOA estimation algorithm is then applied to the optimized sparse towed linear array,resulting in dimensionality reduction acceleration and high-precision DOA estimation,demonstrating its robustness in complex marine environments.Experimental data processing confirms the excellent practical value and engineering application potential of the proposed algorithms. |
