【作者】 曾操；

【导师】 廖桂生；

【作者基本信息】 西安电子科技大学 ， 信号与信息处理， 2008， 博士

【摘要】 论文主要研究多通道地面动目标检测雷达中的稳健阵列处理技术,首先针对传统阵列研究稳健处理共性技术,然后将稳健阵列处理中的共性关键技术应用于多通道地面动目标检测雷达系统以提高地面运动目标的检测性能。论文的主要内容包括以下几个方面:1.针对对角加载自适应波束形成方法中的加载量难以确定问题,提出一种基于子空间投影的加权代价函数迭代搜索出合适加载量的方法。通过构造导向矢量分别向信号子空间和噪声子空间投影的加权代价函数来评价加载量的合适与否。该方法无需给定系统误差范围,通过迭代搜索能够快速、唯一地获得合适加载量,并且获得的加载量对初始值不敏感。线阵和圆阵仿真结果验证了所提方法对输入信噪比变化、指向误差以及幅相误差具有良好的稳健性。2.针对阵列流形误差导致自适应波束形成器输出性能下降的问题,提出一种旁瓣电平软约束条件下搜索人工干扰功率值的稳健波束形成方法。首先,在波达方向粗估基础上,采用期望信号的导向矢量向信号子空间投影和加宽干扰零陷区的技术提高线性约束最小均方误差LCMV波束形成器的稳健性。其次,允许旁瓣最大峰值起伏在一定范围内,通过粗、精两种模式搜索出估计干扰零陷区的注入功率值以兼顾零陷深度和旁瓣电平要求。在相同误差条件下,与传统的对角加载方法、锥消方法以及稳健的Capon波束形成方法相比,仿真结果显示所提方法具有更高的输出信干噪比。3.为实现子阵级多波束和避免由于子阵相位中心均匀稀布带来的栅瓣问题,提出基于对称指数分布的子阵相位中心优化方法和非均匀子阵级窗函数。在阵列孔径和阵元个数给定的情况下,提出选取合适指数参数r的步骤并用于子阵等效相位中心设计,根据子阵中心推导了非均匀泰勒(Taylor)窗函数,最后用多指向的加窗空域匹配权实现子阵级数字多波束。仿真结果显示优化子阵中心的线阵采用提出的非均匀加权具有无栅瓣,近区低旁瓣,阵列效率损失小等优点;应用于对称阵子组成的线阵,获得的方向图结果与数值电磁代码(NEC)软件计算的结果吻合;推广到对称指数分布面阵同样具有有效性。4.针对沿航迹多通道地面动目标检测雷达杂波数据存在的非均匀、样本不够不强以及强运动目标或孤立干扰带来的样本污染问题,提出一种稳健的空域导向矢量估计方法。首先,空域导向矢量估计包括相位和幅度估计两个阶段。采用功率挑选估计出主杂波区中各段的杂波相位矢量矩阵,对矩阵分解得到的归一化最大特征矢量的相位进行回转中值滤波以滤除相位突变,相位解缠绕后拟合出所有方位向的相位估计,进而得到相位导向矢量估计;对估计出的杂波矢量矩阵特征分解后,采用坏点附近取平均的方式消除其影响,获得幅度导向矢量的估计。其次,杂波有效样本通过基于该导向矢量的波束形成被挑出以估计杂波相关矩阵,杂波相关矩阵特征分解后最小特征向量作为自适应杂波相消权矢量。抑制杂波后检测出运动目标,最后通过导向矢量与动目标数据矢量匹配的方式实现定位和测速。实测数据处理结果表明空域导向矢量估计方法能有效滤除相位估计突变和幅度估计坏点;提出的样本挑选方法与基于功率的挑选方法相比具有更好的动目标检测性能;基于空域导向矢量匹配的定位和测速效果良好。5.针对机载三通道雷达系统的杂波抑制问题,提出基于样本加权杂波相关矩阵稳健估计的三通道联合动目标检测方法。基于广义内积值相对距离概念对像素矢量样本加权,三通道联合杂波抑制后检测出动目标初始位置,局部两两干涉实现动目标重定位和测速,最后进行了性能分析。所提方法无需设置样本选取门限,对图像配准精度仅要求在1个像素以内,样本加权后改善因子提高明显;3幅图像联合处理的性能优于2幅图像处理。机载实测数据处理结果验证了方法的有效性和稳健性。更多还原

【Abstract】 This dissertation addresses the robust array processing techniques in the multichannel ground moving targets indicator (GMTI) radars. Firstly, some robust array processing techniques have been studied. Secondly, some mutual and key techniques in robust array processing are applied in the multichannel GMTI radars to improve the performance of detecting the ground moving targets. The main topics of this dissertation are listed as follows:1. Aimming at the problem of how to choose the diagonal loading (DL) level used in the adaptive beamforming methods of diagonal loading, a new method for estimating the appropriate diagonal loading level is proposed based on the weighted cost function of subspace projection. Without the evalution of system error, the appropriate diagonal level can be rapidly and uniquely obtained through an iterative search which is not sensitive to the given initial valve. Simulations demonstrate the robustness under the variation of output signal-to-noise ration (OSNR) to input signal-to-noise ration (ISNR), look direction error and gains and phases error with our method.2. Aimming at the performance degration of the adaptive beamformer caused by the array manifold error, a new robust beamforming method is proposed for estimating the artificial interference power under the sidelobe soft constraint. Firstly, the robustness of linear constraint minimum variance (LCMV) beamformer is improved by the projection of the steering vector of the signal of interest (SOI) into the signal subspace and the extension technique of the interferences area with the imprecise knowledge of direction of arrival (DOA). Secondly, the artificial interference power can be estimated through the iterative procedure under the soft constraint of the maximum peak undulation in sidelobe area, which is used to satify the requirement of null depth and sidelobe level. Numerical examples demonstrate it has a higher performance of output signal-to-interference-plus-noise ration (OSINR) than Traditional Diagonal Loading (TDL), Covariance Matrix Taper (CMT) and Robust Capon beamformer (RCB) under the same error conditions.3. An optimized method for the phase centres of subarray based on the symmetric exponentially distribution and an ununiform window of subarray are proposed to realize the subarray multibeam-forming and avoid the grating lobe due to the uniform sparse phase centers of subarray. Firstly, given the aperture and the number of the array, the phase centers of the subarray are optimized through an approach of choosing the approatiate exponential parameter. Secondly, an ununiform Taylor window is derived for the sidelobe control according to the optimized phase centers. Finally, the subarray multibeam-forming can be achieved with the windowed steering vector pointing at the various directions. Simulation results show that the linear array with the optimized phase centers and the ununiform window has the advantages of no grating lobe, low sidelobe level nearby the mainlobe and the high array efficiency over other methods; The pattern results of our method applied in the linear array composed of symmetric array elements agree with that of Numerical Electromagnetic Code (NEC) software; The proposed method spreading to the plane array of symmetric exponentially distribution is available.4. For the problems of the clutter data obtained by the multichannel GMTI radar system, such as heterogeneity, insufficiency or weakness and contamination due to the strong moving targets or the isolated interferences, a new method for estimating the spatial steering vectors is proposed. Firstly, the spatial steering vectors are accomplished by two stages of phase and amplitude estimation. Secondly, the suitable clutter data, which are picked out by the beamforming mode with the estimated steering vectors, are used to estimate the adaptive weight for clutter rejection. Thirdly, the moving targets are detected with the results of clutter suppression. Finally, the locations and the radial velocities are estimated by matching the estimated steering vector and the data vector of moving targets. The results of the airborne measured data demonstrate that the jump points in the phase estimation and the bad points in the amplitude estimation can be eliminated; the sample selection method using beamforming has better detection performance than that using power order; the moving targets can be relocated to the road using the match method.5. A robust method for estimating the covariance matrix using the weighted pixels vector for the joint clutter rejection is proposed. Firstly, the pixels vector is weighted by the relative range of the generalized inner product. Secondly, the locations of the ground moving targets are detected using the map of joint clutter rejection with three SAR images. Thirdly, the true locations and velocities of the targets can be obtained by the local interferometric method. Finally, the performance analysis is given. Without the threshold of the sample select and only one pixel image registration being required, the improved factor (IF) of the weighted processing has been improved much more than that of the noweighted processing and the IF of the jointed processing is better than that of the paired processing. The effectiveness and robustness of the proposed method are verified with the process results of measured airborne data.更多还原