基于稀疏反演的相参捷变频雷达信号处理

发布时间：2018-08-06 12:15

【摘要】：相参捷变频雷达具有优异的低截获、抗干扰和电磁兼容性能，研究其信号处理问题具有重要的应用价值。论文主要考虑捷变频雷达中多目标距离速度联合估计问题，围绕在信号处理中如何利用观测场景中稀疏信息的问题展开研究，取得了以下成果。推导了捷变频雷达脉冲回波的信号模型，将距离-速度联合估计问题建模为线性方程组求解的问题，并说明该线性方程组是欠定的。采用传统的匹配滤波处理欠定方程组，会出现严重的旁瓣平台，造成虚警、小目标被旁瓣掩盖等问题。由于在同一个粗分辨单元内目标个数往往较少，捷变频雷达观测场景呈现出了明显的稀疏性。压缩感知算法通过挖掘场景的稀疏性这一先验信息，能够有效地求解该线性方程组，从而抑制旁瓣、准确重建场景。论证了压缩感知算法在捷变频雷达中的适用性。通过严格分析捷变频雷达中观测矩阵的性质，定量地给出了关于场景稀疏度和雷达参数的充分条件。当满足该条件时，利用压缩感知算法可在无噪声的情况下精确重建稀疏场景，，或在有噪声的情况下稳健恢复场景。仿真和实测数据也验证了压缩感知算法在捷变频雷达中距离-速度联合估计中的有效性。解决了压缩感知算法中的模型失配问题。实际观测场景通常是在距离速度连续二维空间上稀疏。将连续空间离散化时，场景中目标可能并不在格点上，从而出现模型失配问题，导致传统压缩感知算法性能下降。提出了基于总体最小二乘的自适应匹配追踪算法来解决该失配问题。算法将格点误差建模成未知参数，采用约束总体最小二乘算法自适应地估计格点误差，并利用估计结果相应地调整观测矩阵，降低模型失配的影响，从而提高场景估计的鲁棒性。将认知机制引入捷变频雷达中，利用已经获得的场景先验信息，进一步提高压缩感知算法重建场景的精度。提出了最小化重建误差的Cramer-Rao界(CRB)作为雷达脉冲载频的优化设计准则，通过降低CRB使得雷达观测系统能够提供更多信息，降低不同目标回波之间的相互干扰。论文还根据捷变频雷达观测矩阵的结构特点，提出了最小化CRB的一种近似准则，能够有效地降低计算量。根据雷达不同的潜在需求，提出了序贯和批处理两种优化载频的工作模式。
[Abstract]:Coherent frequency agile radar has excellent performance of low interception, anti-jamming and electromagnetic compatibility, so it has important application value to study the problem of signal processing. In this paper, the problem of joint range velocity estimation of multi-target in frequency-agility radar is considered. The problem of how to use sparse information in observation scene in signal processing is studied, and the following results are achieved. The signal model of pulse echo of agile frequency conversion radar is derived. The range velocity joint estimation problem is modeled as a linear equation system, and it is shown that the linear equation group is under determined. Using traditional matched filtering to deal with underdetermined equations will lead to serious sidelobe platform resulting in false alarm and small target being covered by sidelobe. Because of the small number of targets in the same coarse resolution unit, the observational scene of frequency agile radar shows obvious sparseness. By mining the prior information of scene sparsity, the compressed perceptual algorithm can solve the linear equations effectively, which can suppress the sidelobe and reconstruct the scene accurately. The applicability of compression sensing algorithm in frequency agility radar is demonstrated. By strictly analyzing the properties of observation matrix in frequency agility radar, the sufficient conditions about scene sparsity and radar parameters are given quantitatively. When this condition is satisfied the sparse scene can be accurately reconstructed without noise or the scene can be recovered stably in the presence of noise by using the compression sensing algorithm. The simulation and measured data also verify the effectiveness of the compression sensing algorithm in the range velocity joint estimation of frequency agility radar. The problem of model mismatch in compressed sensing algorithm is solved. The actual observation scene is usually sparse in the range velocity continuous two-dimensional space. When the continuous space is discretized, the target in the scene may not be on the lattice point, which leads to the model mismatch problem, which leads to the performance degradation of the traditional compression sensing algorithm. An adaptive matching tracking algorithm based on population least squares is proposed to solve the mismatch problem. The lattice error is modeled as an unknown parameter, and the constrained population least square algorithm is used to estimate the lattice error adaptively, and the observation matrix is adjusted accordingly using the estimation results to reduce the effect of model mismatch. In order to improve the robustness of scenario estimation. The cognitive mechanism is introduced into the frequency agility radar and the scene prior information obtained is used to further improve the accuracy of the compressed perceptual algorithm to reconstruct the scene. Cramer-Rao bound (CRB), which minimizes the reconstruction error, is proposed as the optimal design criterion for radar pulse carrier frequency. By reducing the CRB, the radar observation system can provide more information and reduce the mutual interference between the echoes of different targets. According to the structural characteristics of the observational matrix of frequency agile radar, an approximate criterion for minimizing CRB is proposed, which can effectively reduce the computational complexity. According to the different potential requirements of radar, two optimal carrier frequency operation modes, sequential and batch processing, are proposed.
【学位授予单位】：清华大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN957.51

【参考文献】