| The Krylov subspace method is applied to radar multi-channel signal detection,which is a relatively new research field.The traditional signal detection method needs to solve the inverse of covariance matrix,which has some difficulties in engineering implementation.As a classical iterative algorithm of krylov subspace,conjugate gradient algorithm can solve the inverse approximate solution of covariance matrix by iteration.In the multi-channel radar signal detection,the covariance matrix has a high number of conditions,and the conjugate gradient algorithm is used to solve the filter weight to converge to the value,which does not meet the actual operation requirements.The pretreatment of conjugate gradient algorithm is based on the precondition of covariance matrix,so that the original covariance matrix condition can be reduced to the maximum extent,and the iterative convergence of conjugate gradient algorithm is accelerated.In this paper,in the process of multi-channel signal detection,the condition of covariance matrix is large and the convergence rate of conjugate gradient algorithm is slow.The preprocessing conjugate gradient algorithm is used to solve the adaptive weight value in the krylov subspace,and the convergence speed of conjugate gradient algorithm is accelerated.The main work of this paper includes:1.Introduced the Krylov subspace conjugate gradient algorithm,and analyzed the principle of the conjugate gradient algorithm to accelerate the iterative convergence of weight vector.Secondly,several classical precondition processing conjugate gradient algorithms are introduced and the advantages and disadvantages of each method are analyzed.Based on the background of the project,a radar signal detection model was established,and the conjugate gradient algorithm matched filter and related properties were introduced.2.Appling Symmetric overrelaxation preconditioning conjugate gradient method to the multi-channel signal detection,get matched filter preconditioning conjugate gradient(PCG-MF)and preconditioning conjugate gradient adaptive matched filter(PCG-AMF).Analysis of symmetrical overrelaxation preconditioning conjugate gradient algorithm iterative weight vector convergence speed and need of iterative computation,verified the algorithm is to accelerate the weight vector iterative convergence and reduce the computational complexity,etc.Based on airborne radar clutter data,the detection performance of PCG-MFand PCG-AMF is better than that of CG-MFand CG-AMFin the same iteration times.3.Appling the preprocessing conjugate gradient to multi-channel parameterized signal detection,get the parameterized adaptive matching filter(PCG-PAMF).According to the generalized stationary characteristics of clutter,a hybrid low-order multi-channel AR model is established.In this paper,the time domain albination parameters of the parameterized matched filter and the spatial albinarization parameters are obtained by using the block cyclic preprocessing conjugate gradient algorithm for the block toplitz characteristics of covariance matrix.Finally,the detection performance of PCG-PAMF was analyzed.4.The conjugate gradient algorithm matching filter is implemented in FPGA.Based on the detection statistics of conjugate gradient algorithm,the weight vector iterative solution module and the detection statistic module of conjugate gradient algorithm are designed.Through the modelsim and matlab simulation tools,the correctness of the realization module is verified. |
