Krylov子空间多通道参数化自适应信号检测方法
发布时间：2018-10-26 10:15
【摘要】：多通道信号检测问题是雷达、通信和医疗等领域的主要研究课题之一,经典的信号检测方法存在的问题主要有全空时处理计算量大,实际非均匀环境中符合独立同分布条件的训练样本较少,由此导致信号检测性能下降以及实时运算困难。本文针对经典多通道信号检测方法中存在的问题,采用Krylov子空间方法迭代地计算空时二维权向量,并利用杂波数据在脉冲域的平稳性,将杂波近似为AR模型。在此基础上,研究Krylov子空间自适应匹配滤波器和Krylov子空间多通道参数化自适应信号检测方法,主要工作及相关结论包括:1、将Krylov子空间方法应用于AMF检测器,并分析迭代过程中产生的一系列KAMF检测器的虚警概率。杂波协方差矩阵具有低秩校正结构形式时,采用共轭梯度法收敛速度快,且迭代至r(10)1次时有较好的检测性能。根据极端Ritz值的快速收敛性、?R-正交投影定理和Wishart分布分析KAMF检测器在各迭代次数下的虚警概率,并给出1?k?r(10)1时的近似理论表达式。基于仿真及实测数据对虚警概率及检测概率作验证,理论分析及仿真结果均说明KAMF检测器的虚警概率介于MF和AMF之间。同时,迭代次数k(28)r(10)1时,检测概率优于AMF。2、将Krylov子空间方法应用于多通道参数化自适应信号检测,采用共轭梯度法解Wiener-Hopf方程,得到一系列的KPAMF检测器。迭代完成时,KPAMF与PAMF的检测性能一致,且多数情况下,KPAMF能在较少的迭代次数内收敛,达到进一步降低运算量的目的。同时,杂波协方差矩阵的条件数较大时,采用预处理的共轭梯度法可降低条件数,提高检测器收敛速度。另一方面,对干扰占主导地位的杂波,AR模型的阶数、参数与干扰个数及干扰参数有关,且预测向量的自相关矩阵具备低秩校正结构。基于AR仿真数据、干扰模型及实测数据论证了以上KPAMF检测器的相关性质及方法。3、对功率谱非均匀和统计分布非均匀的杂波环境建模,并将KAMF检测器和KPAMF检测器应用于非均匀环境中的目标检测。仿真结果表明,由于非均匀环境下有效训练样本不足,KPAMF检测器和PAMF的杂波抑制效果优于KAMF检测器和AMF。同时,KAMF检测器在特定迭代次数下的检测效果优于AMF,KPAMF检测器在较少的迭代次数内接近于PAMF的检测效果,论证了KAMF检测器和KPAMF检测器的相关性质及结论。
[Abstract]:The problem of multi-channel signal detection is one of the main research topics in the fields of radar, communication and medical treatment. In the non-uniform environment, there are fewer training samples which meet the condition of independent and same distribution, which leads to the deterioration of signal detection performance and the difficulty of real-time operation. In order to solve the problems in classical multi-channel signal detection methods, Krylov subspace method is used to iteratively calculate space-time two rights vector, and the clutter is approximated to AR model by using the smoothness of clutter data in pulse domain. On this basis, the Krylov subspace adaptive matched filter and the Krylov subspace multi-channel parameterized adaptive signal detection method are studied. The main work and related conclusions are as follows: 1. The Krylov subspace method is applied to the AMF detector. The false alarm probability of a series of KAMF detectors generated during iteration is analyzed. When the clutter covariance matrix has a low rank correction structure, the conjugate gradient method has a fast convergence rate and good detection performance when iterated to r (10) once. According to the fast convergence of extreme Ritz values,? R- orthogonal projection theorem and Wishart distribution, the false alarm probability of KAMF detector under each iteration is analyzed, and the approximate theoretical expression of 1?k?r (10) 1 is given. The false alarm probability and detection probability are verified based on simulated and measured data. The theoretical analysis and simulation results show that the false alarm probability of KAMF detector is between MF and AMF. At the same time, when the number of iterations k (28) r (10) is 1, the detection probability is better than that of AMF.2,. The Krylov subspace method is applied to multi-channel parameterized adaptive signal detection. The conjugate gradient method is used to solve the Wiener-Hopf equation and a series of KPAMF detectors are obtained. When the iteration is completed, the detection performance of KPAMF and PAMF is the same, and in most cases, KPAMF can converge in a few iterations, so as to further reduce the computational complexity. At the same time, when the condition number of clutter covariance matrix is large, the condition number can be reduced and the convergence rate of detector can be improved by using the conjugate gradient method of preprocessing. On the other hand, for clutter dominated by interference, the order and parameter of AR model are related to the number of disturbances and interference parameters, and the autocorrelation matrix of prediction vector has low rank correction structure. Based on the AR simulation data, interference model and measured data, the related properties and methods of the above KPAMF detectors are demonstrated. 3. The clutter environment with non-uniform power spectrum and statistical distribution is modeled. KAMF detector and KPAMF detector are applied to target detection in non-uniform environment. The simulation results show that the clutter suppression effect of KPAMF detector and PAMF is better than that of KAMF detector and AMF. because of the shortage of effective training samples in non-uniform environment. At the same time, the detection effect of KAMF detector is better than that of AMF,KPAMF detector in less iterations than that of PAMF detector under certain iterations. The properties and conclusions of KAMF detector and KPAMF detector are proved.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN911.23
本文编号：2295389
[Abstract]:The problem of multi-channel signal detection is one of the main research topics in the fields of radar, communication and medical treatment. In the non-uniform environment, there are fewer training samples which meet the condition of independent and same distribution, which leads to the deterioration of signal detection performance and the difficulty of real-time operation. In order to solve the problems in classical multi-channel signal detection methods, Krylov subspace method is used to iteratively calculate space-time two rights vector, and the clutter is approximated to AR model by using the smoothness of clutter data in pulse domain. On this basis, the Krylov subspace adaptive matched filter and the Krylov subspace multi-channel parameterized adaptive signal detection method are studied. The main work and related conclusions are as follows: 1. The Krylov subspace method is applied to the AMF detector. The false alarm probability of a series of KAMF detectors generated during iteration is analyzed. When the clutter covariance matrix has a low rank correction structure, the conjugate gradient method has a fast convergence rate and good detection performance when iterated to r (10) once. According to the fast convergence of extreme Ritz values,? R- orthogonal projection theorem and Wishart distribution, the false alarm probability of KAMF detector under each iteration is analyzed, and the approximate theoretical expression of 1?k?r (10) 1 is given. The false alarm probability and detection probability are verified based on simulated and measured data. The theoretical analysis and simulation results show that the false alarm probability of KAMF detector is between MF and AMF. At the same time, when the number of iterations k (28) r (10) is 1, the detection probability is better than that of AMF.2,. The Krylov subspace method is applied to multi-channel parameterized adaptive signal detection. The conjugate gradient method is used to solve the Wiener-Hopf equation and a series of KPAMF detectors are obtained. When the iteration is completed, the detection performance of KPAMF and PAMF is the same, and in most cases, KPAMF can converge in a few iterations, so as to further reduce the computational complexity. At the same time, when the condition number of clutter covariance matrix is large, the condition number can be reduced and the convergence rate of detector can be improved by using the conjugate gradient method of preprocessing. On the other hand, for clutter dominated by interference, the order and parameter of AR model are related to the number of disturbances and interference parameters, and the autocorrelation matrix of prediction vector has low rank correction structure. Based on the AR simulation data, interference model and measured data, the related properties and methods of the above KPAMF detectors are demonstrated. 3. The clutter environment with non-uniform power spectrum and statistical distribution is modeled. KAMF detector and KPAMF detector are applied to target detection in non-uniform environment. The simulation results show that the clutter suppression effect of KPAMF detector and PAMF is better than that of KAMF detector and AMF. because of the shortage of effective training samples in non-uniform environment. At the same time, the detection effect of KAMF detector is better than that of AMF,KPAMF detector in less iterations than that of PAMF detector under certain iterations. The properties and conclusions of KAMF detector and KPAMF detector are proved.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN911.23
【参考文献】
相关期刊论文 前5条
1 吴洪;王永良;陈建文;;线性预测类STAP方法研究[J];电子与信息学报;2008年05期
2 任双桥;刘永祥;黎湘;庄钊文;;广义相关K分布杂波建模与仿真[J];自然科学进展;2006年06期
3 丁前军,王永良,张永顺;降秩自适应滤波算法研究[J];雷达科学与技术;2005年04期
4 方学立,梁甸农;一种用于相关高斯序列仿真的新方法[J];雷达科学与技术;2004年04期
5 汤俊,彭应宁;共轭梯度自适应波束形成算法[J];电子科学学刊;2000年01期
相关博士学位论文 前1条
1 唐斌;机载雷达Krylov子空间STAP算法研究[D];电子科技大学;2008年
，本文编号：2295389
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2295389.html
