Research On Radar Sparse Imaging Method Based On Multi-objective Evolutionary Algorithm

High resolution imaging is a research hotspot in signal processing field.High resolution one-dimensional range profile of radar target is widely used in target recognition,air traffic control and other fields.In recent years,the compressed sensing method proposed by researchers can obtain high resolution range profile through a small amount of observation data.However,the current compressed sensing algorithm mainly has two problems: first,it is difficult to set regularization parameters;second,it is easy to fall into local optimal solutions.It is an effective solution to transform the sparse recovery problem of compressed sensing into multi-objective optimization problem.Firstly,the solution of the multi-objective optimization problem does not depend on the setting of regularization parameters.Secondly,the solution of the multi-objective optimization problem is global search,and its solution is a set of optimal solutions.Multi-objective evolutionary algorithm performs better in solving multi-objective optimization problems.In radar imaging,the full polarization processing can extract the coherent polarization information of the scattering target,and the optimal polarization channel obtained by using the full polarization information can improve the precision of selffocusing imaging.The traditional compressed sensing method lacks full polarization information.Therefore,in this paper,the full polarization processing model of parametric sparse recovery is established around the radar target high resolution range profile,and the full polarization radar imaging algorithm based on multi-target evolutionary algorithm is studied.The main contents of this paper are as follows:1.The basic theory and common algorithms of radar target sparse imaging and multi-target evolution are summarized.In this paper,the principles of sparse imaging methods of greedy tracking class,convex optimization class and statistical method class are reviewed,and the representative algorithms are selected for simulation experiments,and the performance of these algorithms is analyzed and compared.The basic framework of multi-objective evolutionary algorithm is studied,and the basic idea and algorithm flow of NSGA-ⅱ and MOEA/D multi-objective evolutionary algorithm are given.2.Radar imaging method based on multi-objective evolutionary algorithm is studied.Firstly,a joint sparse recovery parameterization model with random stepping frequency signals is established.Then,the parametric sparse recovery problem is transformed into a two-objective optimization problem,and the joint estimation of motion parameters and high resolution range profile is realized by using multi-objective evolutionary algorithm.The multi-objective evolutionary algorithm used in this paper is MOEA/D algorithm,which mainly adopts a multi-stage search method.3.A multi-objective optimization model with two kinds of objectives is established by combining full polarization treatment with parametric sparse recovery: parametric joint sparse recovery optimization target and full polarization treatment optimization target.In the end,not only can range direction and motion parameters be estimated,but also polarization information such as optimal polarization channel can be obtained.