Array Sparse Optimization And DOA Estimation

In recent years,the demand for design of radar detection systems based on phased array systems has increased.Considering the maintenance of hardware equipment and resource consumption,how to select finite array elements among multiple array elements for signal estimation has become a research hotspot.However,in practical applications,the existing sparse array structure still has the problems of limited array element spacing.Therefore,the purpose of sparse array optimization design is to find an optimization method that can recognize signals with fewer elements under restricted conditions and has better direction of arrival(DOA)estimation accuracy.In the field of DOA estimation,meshless DOA estimation has received high attention due to its better estimation performance in the case of grid mismatch.Considering that the existing meshless DOA estimation algorithms are mainly suitable for two-dimensional planar arrays with regular physical array elements such as uniform rectangular arrays and sparse rectangular arrays,the DOA estimation problem that is suitable for optimized arbitrary planar arrays remains to be solved.In order to solve these problems,this article mainly proposes innovations from two aspects:array optimization method and DOA estimation algorithm.Firstly,in view of the high cost of the existing uniform arrays,and the current common sparse arrays cannot meet the array aperture constraints under actual conditions,an array optimization design based on adaptive genetic algorithm is proposed.Starting from the two aspects of array performance constraints and array aperture constraints,intelligent algorithms are used to optimize the constraints of the array.Finally,the simulation verifies the effectiveness of the algorithm and the DOA estimation performance of the array optimized by the algorithm.Secondly,in order to solve the problems that the intelligent optimization algorithm may fall into the local optimization value prematurely and the amount of calculation is large,an array optimization algorithm based on reinforcement learning is proposed.The reinforcement learning algorithm is used to enable the agent to learn correctly under the condition of less information in the environment,and to meet the advantages in the dynamic unknown environment,while satisfying the constraints of the array aperture,using the Cramerlo bound and the degree of freedom as the reward value.Array optimization.Simulations verify the effectiveness of the algorithm and the DOA estimation performance of the array obtained by the algorithm optimization.Finally,the performance of arrays optimized based on adaptive genetic algorithm and reinforcement learning algorithm are compared,and their advantages and disadvantages are explained.Finally,we proposed a meshless DOA estimation algorithm based on atomic norm which is suitable for arbitrary planar arrays.We use a kind of Bessel function to expand the covariance data of the received data of the arbitrary geometric structure plane array,and then transform the expanded matrix into an Semi-Definite Programming(SDP)problem.Through the Vandermonde decomposition of the positive semi-definite matrix,the corresponding parameter transformation is carried out.Obtain the incident angle of the source and complete the twodimensional DOA estimation.The effectiveness of the algorithm is verified by simulating the incident signal source of any array,and the direction finding performance is compared with the Multiple Signal Classification(MUSIC)algorithm that is also applicable to any array,and the accuracy of the algorithm is verified.