稀疏阵列测向技术研究
发布时间：2019-04-10 12:24
【摘要】：稀疏阵是指阵元间距大于半波长的阵列系统。与常规满阵相比,稀疏阵用相同数量的阵元能获得更大的阵列孔径,甚至形成更多的虚拟阵元,因而它具有许多优越的测向性能,如较强的分辨能力、较高的估计精度和较大的信源处理能力,这些优势使之成为当前阵列测向方面的研究热点。论文以稀疏阵列为研究对象,研究稀疏阵列结构设计相应的波达角(DOA)估计算法及其统计性能,从理论和仿真上定量地证明稀疏阵列在测向方面的优势和潜能。论文主要工作如下： 1.研究了多级互质阵列的解模糊容差及其扩展孔径测向算法。首先推导了三种解模糊算法的解模糊容差表达式,定量地证明了多级互质阵列具有形成较大解模糊容差的能力,并通过外场实验验证了理论的正确性。然后根据这种阵列的几何结构,分别提出一种模转换虚拟旋转不变因子(MC-VESPRIT)法和一种模转换虚拟传播因子(MC-VPM)法。这两种算法都利用阵元间距之间的多级互质关系,以及四阶累积量的阵列扩展特性,以突破传统算法对参考阵元间距的半波长限制,从而有效地提高了测向精度。更为重要地是,MC-VPM算法还采用双L型结构避免了角度估计失败,并利用VPM算法获得了自动配对的二维(2-D) DOA估计。 2.研究了两种嵌套阵的设计及其2-D DOA估计算法。首先,设计了一种包含多尺度阵元间距的双十字型嵌套阵,以期获得自由度的增加和两次阵列孔径扩展。为了实现这一目的,提出一种虚拟矩阵束(VMPM)算法用于2-D DOA估计。VMPM算法利用多尺度阵元间距,构造了由更多虚拟传感器组成的双平行阵。与传播因子(PM)改进算法相比,它也使用二阶统计量,却拥有更高的测角精度和处理O(P2/32)个信号源的能力,其中P为传感器个数。另外,还分析了VMPM算法和PM改进算法的统计性能,推导了它们估计误差的渐近方差表达式和克拉美罗下界(CRLB),并将这些表达式简化为单信号情况,以便定量地证明VMPM算法的性能优势。接着,将上述嵌套思想拓展到多级,设计了一种多级嵌套L型阵,并提出一种基于二维空间平滑的2q多重信号分类(2q-MUSIC)算法。该算法利用2q阶统计量和M+N个物理阵元,系统地形成了有O(Mq×Nq)个虚拟阵元的均匀矩形阵(URA),进一步提高了测向性能。此外,还推导了多级嵌套L型阵的最优和次优配置表达式,以便最大化虚拟URA的阵元数。 3.研究了三种电磁矢量稀疏阵的设计及其2-D DOA估计算法。首先,对互质／嵌套标量阵在二维空间-极化联合域进行推广,设计了电磁矢量互质／嵌套面阵。这两种面阵除了在空域蕴含二维互质／嵌套特性外,还在极化域蕴含多样性。通过充分挖掘这三维特性,形成包含更多自由度的差合成电磁矢量URA,从而获得自由度的增加和两次孔径扩展。为了将这些优势用于2-D DOA估计,以及将传感器恢复成6分量结构,提出一种基于三维平滑的极化多重信号分类(3DS-PMUSIC)算法。相比于PMUSIC算法,该算法在测向精度、分辨率和最大可处理信源数等方面都有较大地改善,其中以最大可处理信源数方面最为突出。理论和仿真结果都表明其最大可处理信源数可以从O(P)提高到O(3P2)。之后,针对空间相关噪声背景下2-D DOA和极化参数估计性能下降的问题,提出一种基于三尺度平行矢量阵的矩阵重构算法。该算法通过构造特殊的互相关矩阵,不仅去除了噪声项,而且充分挖掘了阵列的空间-极化特性,因而获得估计性能的提高。另外,该算法无需矩阵开方和高阶统计量,具有低运算量。 4.针对分布式电磁矢量稀疏线阵,提出一种基于增强矩阵的测向算法和另一种基于PM-矩阵重构的参数估计算法。其中,前者用于估计相干信号的2-D DOA,后者用于估计混合信号(相干信号和独立信号共存)的2-D DOA和极化参数。两者不但都保留了传感器矢量特性,而且允许传感器内部分量的间隔和传感器之间的间隔可以超过半波长,因而测向精度较高。尤其是后者,它还将独立信号和相干信号分开处理,使得阵列孔径得到更加充分地利用,进而改善测向精度和最大可处理信源数,甚至避免了独立信号和相干信号因入射角相近时而导致的估计性能下降问题。另外,针对上述阵列,还推导了混合信号2-D DOA和极化参数联合估计的CRLB。
[Abstract]:Sparse array is an array system with array element spacing greater than half wavelength. Compared with the conventional full array, the sparse array can obtain larger array aperture and even more virtual array elements by the same number of elements, so that it has many superior direction finding performance, such as higher resolution capability, higher estimation precision and larger source processing capability, These advantages make it a hot spot in current array direction finding. In this paper, a sparse array is used as the research object to study the DOA estimation algorithm and its statistical performance of the sparse array structure design, and the advantages and potential of the sparse array in direction finding are quantitatively proved from the theory and the simulation. The main work of the thesis is as follows: 1. The solution fuzzy tolerance of multi-stage inter-quality array and its extension aperture direction-finding calculation are studied. In this paper, we first derive the solution-fuzzy tolerance expression of three de-fuzzy algorithms, and quantitatively prove that the multi-level mutual-quality array has the capability of forming a large solution-fuzzy tolerance, and the correctness of the theory is verified by the field experiment. And then, according to the geometrical structure of the array, a mode conversion virtual rotation invariant factor (MC-VESPRIT) method and a mode conversion virtual propagation factor (MC-VPM) are respectively provided. The method comprises the following steps: using the multi-level mutual-quality relation between the array element spacing and the array extension characteristic of the fourth-order cumulant, so as to break through the half-wavelength limitation of the distance between the reference array elements by the traditional algorithm, thereby effectively improving the direction-finding precision; Degree. More importantly, the MC-VPM algorithm avoids the failure of the angle estimation by adopting the double-L type structure, and obtains the two-dimensional (2-D) DOA estimation of the automatic pairing by using the VPM algorithm. 2. The design of two nested matrices and its 2-D DOA estimation are studied. In this paper, a double-cross-type nested matrix with multi-scale array elements is designed, with a view to obtaining an increase in freedom and two array holes. To achieve this, a virtual matrix bundle (VMPM) algorithm is proposed for 2-D DO A. The VMPM algorithm uses the multi-scale array element spacing to construct a double-pair of more virtual sensors. Parallel matrix. Compared with the propagation factor (PM) improvement algorithm, it also uses the second order statistics, but has higher measurement accuracy and the ability to process the O (P2/32) signal source, where P is the sensor. In addition, the statistical properties of the VMPM algorithm and the PM improved algorithm are also analyzed. The asymptotic variance and the lower bound (CRLB) of the estimation errors are derived, and these expressions are simplified to a single signal, so as to quantitatively prove the performance of the VMPM algorithm. And then, expanding the nesting thought to a plurality of stages, designing a multi-level nested L-shaped array, and proposing a 2-q multi-signal classification (2q-MUSIC) based on a two-dimensional space smoothing. The algorithm uses the 2 q-order statistic and the M + N physical array elements to form a uniform rectangular array (URA) with O (Mq, Nq) virtual array elements. Performance. In addition, the optimal and sub-optimal configuration expressions for multi-level nested L-matrices are derived in order to maximize the virtual URA 3. The design of three kinds of electromagnetic vector sparse array and its 2-D DO are studied. A. First, the mutual quality/ nested scalar array is generalized in the two-dimensional space-polarization joint domain, and the mutual quality of the electromagnetic vector is designed. In addition to that two-dimensional intertexture/ nesting feature in the spatial domain, the planar area array is also in the polarization domain. with diversity. By fully excavating the three-dimensional characteristics, a difference synthesis electromagnetic vector URA with more degrees of freedom is formed, thereby obtaining an increase in the degree of freedom and two Sub-aperture expansion. In order to use these advantages for 2-D DOA estimation, and to restore the sensor to a 6-component structure, a three-dimensional smooth-based polarization multiple signal classification (3DS-PMUS) is proposed Compared with the PMUSIC algorithm, the algorithm has a great improvement in the direction-finding precision, resolution and the maximum number of processing information sources, in which the maximum number of processing sources The surface is the most prominent. Both the theoretical and the simulation results show that the maximum number of processing sources can be increased from O (P) to O (3). After that, a three-scale parallel vector matrix moment is proposed for the degradation of 2-D DOA and polarization parameter estimation in the background of space-related noise. This algorithm not only removes the noise, but also fully excavates the spatial-polarization characteristics of the array, thus obtaining the estimation. In addition, the algorithm does not need matrix and higher order statistics, 4. Aiming at the sparse linear array of the distributed electromagnetic vector, a direction finding algorithm based on the enhancement matrix and another based on the PM-matrix reconstruction are proposed. The former is used to estimate the 2-D DOA of the coherent signal, which is used to estimate the 2-D D of the mixed signal (the coherent signal and the independent signal co-exist). both the oa and the polarization parameters. both retain the sensor vector characteristics and allow the spacing of the sensor's internal components and the spacing between the sensors to exceed the half-wavelength, in particular the latter, the independent signal and the coherent signal are separately processed, so that the array aperture is more fully utilized, large number of processing sources, even avoiding the estimation of the independent signal and the coherent signal due to the close angle of incidence In addition, for that above-mentioned array, the combination signal 2-D DOA and the polarization parameter combination are also derived.
【学位授予单位】：南京理工大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN911.7
本文编号：2455797
[Abstract]:Sparse array is an array system with array element spacing greater than half wavelength. Compared with the conventional full array, the sparse array can obtain larger array aperture and even more virtual array elements by the same number of elements, so that it has many superior direction finding performance, such as higher resolution capability, higher estimation precision and larger source processing capability, These advantages make it a hot spot in current array direction finding. In this paper, a sparse array is used as the research object to study the DOA estimation algorithm and its statistical performance of the sparse array structure design, and the advantages and potential of the sparse array in direction finding are quantitatively proved from the theory and the simulation. The main work of the thesis is as follows: 1. The solution fuzzy tolerance of multi-stage inter-quality array and its extension aperture direction-finding calculation are studied. In this paper, we first derive the solution-fuzzy tolerance expression of three de-fuzzy algorithms, and quantitatively prove that the multi-level mutual-quality array has the capability of forming a large solution-fuzzy tolerance, and the correctness of the theory is verified by the field experiment. And then, according to the geometrical structure of the array, a mode conversion virtual rotation invariant factor (MC-VESPRIT) method and a mode conversion virtual propagation factor (MC-VPM) are respectively provided. The method comprises the following steps: using the multi-level mutual-quality relation between the array element spacing and the array extension characteristic of the fourth-order cumulant, so as to break through the half-wavelength limitation of the distance between the reference array elements by the traditional algorithm, thereby effectively improving the direction-finding precision; Degree. More importantly, the MC-VPM algorithm avoids the failure of the angle estimation by adopting the double-L type structure, and obtains the two-dimensional (2-D) DOA estimation of the automatic pairing by using the VPM algorithm. 2. The design of two nested matrices and its 2-D DOA estimation are studied. In this paper, a double-cross-type nested matrix with multi-scale array elements is designed, with a view to obtaining an increase in freedom and two array holes. To achieve this, a virtual matrix bundle (VMPM) algorithm is proposed for 2-D DO A. The VMPM algorithm uses the multi-scale array element spacing to construct a double-pair of more virtual sensors. Parallel matrix. Compared with the propagation factor (PM) improvement algorithm, it also uses the second order statistics, but has higher measurement accuracy and the ability to process the O (P2/32) signal source, where P is the sensor. In addition, the statistical properties of the VMPM algorithm and the PM improved algorithm are also analyzed. The asymptotic variance and the lower bound (CRLB) of the estimation errors are derived, and these expressions are simplified to a single signal, so as to quantitatively prove the performance of the VMPM algorithm. And then, expanding the nesting thought to a plurality of stages, designing a multi-level nested L-shaped array, and proposing a 2-q multi-signal classification (2q-MUSIC) based on a two-dimensional space smoothing. The algorithm uses the 2 q-order statistic and the M + N physical array elements to form a uniform rectangular array (URA) with O (Mq, Nq) virtual array elements. Performance. In addition, the optimal and sub-optimal configuration expressions for multi-level nested L-matrices are derived in order to maximize the virtual URA 3. The design of three kinds of electromagnetic vector sparse array and its 2-D DO are studied. A. First, the mutual quality/ nested scalar array is generalized in the two-dimensional space-polarization joint domain, and the mutual quality of the electromagnetic vector is designed. In addition to that two-dimensional intertexture/ nesting feature in the spatial domain, the planar area array is also in the polarization domain. with diversity. By fully excavating the three-dimensional characteristics, a difference synthesis electromagnetic vector URA with more degrees of freedom is formed, thereby obtaining an increase in the degree of freedom and two Sub-aperture expansion. In order to use these advantages for 2-D DOA estimation, and to restore the sensor to a 6-component structure, a three-dimensional smooth-based polarization multiple signal classification (3DS-PMUS) is proposed Compared with the PMUSIC algorithm, the algorithm has a great improvement in the direction-finding precision, resolution and the maximum number of processing information sources, in which the maximum number of processing sources The surface is the most prominent. Both the theoretical and the simulation results show that the maximum number of processing sources can be increased from O (P) to O (3). After that, a three-scale parallel vector matrix moment is proposed for the degradation of 2-D DOA and polarization parameter estimation in the background of space-related noise. This algorithm not only removes the noise, but also fully excavates the spatial-polarization characteristics of the array, thus obtaining the estimation. In addition, the algorithm does not need matrix and higher order statistics, 4. Aiming at the sparse linear array of the distributed electromagnetic vector, a direction finding algorithm based on the enhancement matrix and another based on the PM-matrix reconstruction are proposed. The former is used to estimate the 2-D DOA of the coherent signal, which is used to estimate the 2-D D of the mixed signal (the coherent signal and the independent signal co-exist). both the oa and the polarization parameters. both retain the sensor vector characteristics and allow the spacing of the sensor's internal components and the spacing between the sensors to exceed the half-wavelength, in particular the latter, the independent signal and the coherent signal are separately processed, so that the array aperture is more fully utilized, large number of processing sources, even avoiding the estimation of the independent signal and the coherent signal due to the close angle of incidence In addition, for that above-mentioned array, the combination signal 2-D DOA and the polarization parameter combination are also derived.
【学位授予单位】：南京理工大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN911.7
【参考文献】
相关期刊论文 前10条
1 龚晓峰;徐友根;刘志文;;四四元数域低秩逼近及其在矢量阵列波达方向估计中的应用[J];北京理工大学学报;2008年11期
2 李勇;赵国伟;李滔;;一种机载单站相位干涉仪解模糊算法[J];传感技术学报;2006年06期
3 张聪;邱鹏宇;卢焕章;;基于最小冗余线阵的二维DOA估计方法[J];传感技术学报;2009年03期
4 李博;孙超;;基于最小冗余线阵的阵列扩展方法[J];传感技术学报;2011年03期
5 薛志宏;李广云;卢志辉;储鹄;;高精度测向天线阵基线方位标定方法[J];测绘通报;2007年09期
6 顾建峰;魏平;;基于伪数据域的最小冗余线阵测向算法[J];电波科学学报;2007年06期
7 曲志昱;司锡才;;基于虚拟基线的宽带被动导引头测向方法[J];弹箭与制导学报;2007年04期
8 司伟建;孙圣和;唐建红;;基于阵列扩展解模糊方法研究[J];弹箭与制导学报;2008年04期
9 叶中付;基于约束最小冗余线阵的ESPRIT测向方法[J];电子学报;1997年04期
10 徐友根,刘志文;基于累积量的极化敏感阵列信号DOA和极化参数的同时估计[J];电子学报;2004年12期
相关博士学位论文 前2条
1 顾陈;高分辨阵列信号处理方法研究[D];南京理工大学;2009年
2 张裕峰;多径传播条件下的波达方向估计算法研究[D];中国科学技术大学;2010年
，本文编号：2455797
本文链接：https://www.wllwen.com/kejilunwen/wltx/2455797.html
