【作者】 张斌；

【导师】 王伟；

【作者基本信息】 哈尔滨工程大学 ， 控制科学与工程， 2019， 硕士

【摘要】 雷达成像技术是雷达发展历史上的重要组成部分。近年来兴起的MIMO雷达稀疏成像技术,将MIMO雷达体制和压缩感知技术有机结合起来,可以实时地对感兴趣区域进行高精度成像,因此受到了学术界的广泛关注。然而,将现有的经典贪婪重构算法直接应用于MIMO雷达稀疏成像时,会存在一些问题。比如,经典的贪婪算法支撑集选择和更新策略并不能保证得到高精度的且没有虚假目标的雷达图像;另一方面,当实际目标偏离网格点时,经典的稀疏重构算法的重构性能会急剧下降。针对上述问题,本文主要研究如何构建新的支撑集选择更新策略以及在网格失配情况下的稀疏重构问题。主要工作如下:推导了单基地共址MIMO雷达的稀疏成像模型,从分析空间谱填充大小的角度推导了MIMO雷达成像系统的二维极限分辨率;详细介绍了压缩感知基本原理以及目前主流的稀疏重构算法;最后,确立了雷达系统的参数和成像评价指标。研究了基于改进型贪婪算法的MIMO雷达稀疏成像方法。针对传统贪婪算法存在的伪影和稀疏重构图像分辨率低的问题,提出了基于改进型贪婪算法的混合匹配追踪算法(Hybrid Matching Pursuit,HMP),在算法迭代过程中重新规划了列原子的选择和更新策略,在保证高精度的列原子分辨的基础上又能通过不断的迭代纠正错误的支撑集,具有成像精度高、稳定无伪影的优点。数值仿真结果验证了所提算法的有效性。研究了网格失配(Off-Grid,OG)情况下的稀疏成像问题。首先细致分析了离网问题产生的原因,并且分析了网格细化与网格失配误差之间的关系。在借鉴已有的采用网格细化减缓失配问题思路的基础上,提出了分层聚焦-正交匹配追踪算法(Hierachical Focus Orthogonal Matching Pursuit,HFOMP)。结合支撑集搜寻和分层聚焦的方法,使用外层循环搜索目标散射点的粗坐标,使用内层循环分层细化的方式得到散射点的精确坐标值,而散射系数由校正后的列原子进行估计。与直接加密网格不同的是,所提算法并不会增大测量矩阵的维度。仿真结果证明了所提算法能够较好的克服离网问题的影响并且能够稳健的提取出目标的位置信息和散射强度信息。更多还原

【Abstract】 Radar imaging technology plays an important role in the history of radar development.MIMO radar imaging technology based on compressed sensing has received increasing attention because the technology can accurately image the region of interest in real time,and this technology is also known as MIMO radar sparse imaging technology.However,there are some problems when the existing classic greedy reconstruction algorithm is directly applied to MIMO radar sparse imaging.For example,the classic greedy algorithm support set selection and update strategy does not guarantee high-precision radar images without false targets;on the other hand,when the actual target deviates from the grid point,the reconstruction performance of the classical sparse reconstruction algorithm will drops dramatically.In the framework of MIMO radar sparse imaging,this thesis carries out the following work for the defects of the existing sparse reconstruction algorithm:The sparse imaging model of colocated MIMO radar is derived.The two-dimensional resolution of the MIMO radar imaging system is derived according to the size of the spatial spectral domain filling region.The basic principle of compressed sensing and the current mainstream sparse reconstruction algorithms are introduced in detail as well.Finally,the imaging evaluation parameters of the radar system are selected.All the work laid the foundation for the follow-up research content.The MIMO radar sparse imaging method based on improved greedy algorithm is studied.Traditional greedy reconstruction algorithms have drawbacks in MIMO radar sparse imaging.Aiming at the drawback of traditional greedy algorithm in sparse imaging,such as radar imaging results with artifacts and lower imaging resolution,the Hybrid Matching Pursuit(HMP)algorithm is proposed.The selection and update strategy of column atoms in the iterative process of algorithm is re-planned to ensure high precision.Based on the atomic resolution of the column,it can correct the wrong support set through continuous iteration,and has the advantages of high imaging precision and stability without artifacts.Numerical simulation results verify the effectiveness of the proposed algorithm.The sparse imaging problem with Off-Grid(OG)condition has been addressed in depth.Firstly,the cause of the Off-Grid problem is analyzed in detail,and the relationship between mesh refinement and grid mismatch error is demonstrated as well.Based on the existing idea of using mesh refinement to alleviate the mismatch problem,a Hierarchical Focus-Orthogonal Matching Pursuit Algorithm(Hierachical Focus-OMP,HFOMP)is proposed.Combining the support set search and hierachical focusing method,the outer loop is used to search the coarse coordinates of the target scattering point,and use the inner loop to get the exact coordinates of the scattering points in a layered and refined way.The estimation of the scattering coefficient is the same as the OMP algorithm and is obtained by the least squares method.Unlike the direct refining grid,the proposed algorithm does not increase the dimensions of the measurement matrix,and the recovery accuracy is not affected by the refinement factor.The simulation results show that the proposed algorithm can robustly extract the position information and Radar Cross Section(RCS)of the target under the condition of Off-Grid.更多还原