【作者】 李南君；

【导师】 魏平；

【作者基本信息】 电子科技大学 ， 信号与信息处理， 2010， 博士

【摘要】 阵列测向技术经过四十多年广泛的研究得到了大量的成果。其基本理论架构和算法日趋成熟,但是依然有许多问题尚未解决,原来的经典算法大部分是在普遍的意义上进行测向,而目前随着信号与天线理论的发展,信号和天线结构表现出很强的应用与结构特征,比如宽带、低信噪比、某些特殊的天线结构等。对这些信号源以及天线几何结构特征的应用会对信号测向技术有很大的影响,因此对于一些信号源与天线几何结构特征的充分利用成为当前研究的一大热点。另外阵列信号个数估计技术是高分辨阵列测向技术的一个重要组成部分。实际中,对信号个数的正确估计关系到测向技术的有效使用。随着现在阵列系统越来越接近实际应用,阵列信号个数估计技术逐渐成为一个研究热点。本论文结合上面的研究热点深入探讨了其中的一些问题,获得不少研究成果。主要工作包括研究了低信噪比信号个数估计、宽带信号个数估计与测向方法、二维波达方向估计中的配对问题,阵列扩展以及快速算法问题、如何利用特定的阵列结构和信号源时域特征来提升算法的性能。并进行了大量的仿真实验和理论分析。主要创新之处概括如下:针对高分辨阵列信号个数估计算法进行了详细的研究,利用多通道数据联合估计的思想改进了一种盲波束信号个数估计技术。与原算法相比该改进算法提升了在低信噪比以及不等功率信号入射时的性能,最后的仿真表明改进算法比原算法具有更好的性能,尤其是在低信噪比不等功率信号入射时这个优势更加明显。针对宽带的高分辨阵列处理进行了深入研究,提出了一种具有阵列扩展的宽带信号个数估计与测向方法。该方法主要通过宽带信号各子带对应的波长不同,有意识的构造伪相关矩阵。其信号个数估计方法具有普遍意义。因为从数学上看该方法主要是对盖氏圆盘法做了改进,使噪声与信号圆盘分开的技术适用于非Hermite的Toeplitz矩阵。而现在广泛使用的利用特定阵列结构和数据特征来构造的伪相关矩阵有些是非Hermite的Toeplitz矩阵,同时专门研究对这类矩阵信号个数估计的文献还很少,一般都是假设已知的。但是本文提出算法可以直接用于这一大类矩阵,或者稍加修改即可用于这类矩阵。弥补这方面的一些不足。其测向部分具有很强的阵列扩展性,信号带宽越宽扩展性越强。最后通过一个半实物实验对这个算法做了一些验证。针对L阵二维来波方向估计算法进行了深入的研究,主要研究了天线的物理结构与时空数据的如何利用。创新成果包括结合天线的物理结构特性提出了一种快速算法,该算法主要通过利用伪相关矩阵的降维构造来达到快速运算的目的,同时这种算法还具有自动配对的特点。另外在L阵的基础上结合天线物理结构与信号彼此之间的关系,提出了一种具有自动配对同时兼顾阵列扩展的算法,该算法使用Kronecker积,利用一些数学推导对矩阵进行有意识的重排,从而获得自动配对与阵列扩展的优点。同时还结合特定的阵列结构与信号相关时间比噪声相关时间长这一特点,针对时间色噪声提出了具有阵列扩展特性的算法,由于天线的物理结构信息的引入,该算法还具有自动配对的能力。另外上述三种算法的精度在同类算法表现优异。更多还原

【Abstract】 A number of the superresolution direction-finding (DF) techniques have been proposed during 40 years and the basic theoretical framework has been formed. However, there are still many issues to be solved. Most of classic algorithms only pay attention to estimate DF under general situations. At present, with the development of the signal and antenna techniques, the signal and the antenna structure are showing a very strong application and structural characteristics, Such as wideband, signal-to-noise ratio (SNR), some special antenna structure and so on. The application of the characteristics of the antenna geometry structure and the signal sources will have a great effect to the superresolution DF techniques. Therefore, how to take full advantage of some signal sources and antenna geometry features is a hot topic in current research for DF techniques. Additionally, Sources number estimation is one of the important steps for exploiting DF techniques. The success of DF strongly depends on the correct estimation of sources number. So estimating the number of signals is also a hot topic of high resolution array direction finding technology.Along the hot topics mentioned above, this dissertation carries in-depth research to obtain a lot of results. The main work of this dissertation include studying the estimation of the sources number at low SNR, the sources number and DOA estimation for wideband signals, pair-matching, fast algorithm, array extension in two-dimensional (2-D) DOA estimation, how to make use of the structure of the array sensors together with the data properties for improving the performance of algorithm. The estimation performance of these methods is demonstrated, and the theoretical analysis is confirmed through numerical examples. The main innovation and creation are concluded as follows:Studying the array-based high resolution sources number estimation techniques. Using multi-channel data joint estimation idea to modify a sources number estimation technique based on blind beamforming methods. The simulation shows that the improved algorithm has better performance than the original algorithm, especially in low SNR and incident of unequal power signals.Studying the sources number estimation and DOA techniques for wideband signals. A novel algorithm with array extension properties is proposed. The proposed constructs pseudo-correlation matrix from this property, which is the each sub-narrowband in wideband signal corresponding to the different wavelengths. The sources number estimation part of the algorithm is of universal significance. Because from the mathematical point of view, the mainly idea of this method is to improve the Gerschgorin disk method, adapting the Gerschgorin disk method to toeplitz matrix without the hermite property. Owen to the current widely exploitation of the technology on constructing pseudo-correlation matrix based on making use of the features of the array structure and data, some of which are toeplitz matrix without the hermite property, this method could or make a little modification adapt this situation. So it makes up for some shortcomings in this area. The DOA estimation part of the algorithm has a strong array extension capability, the wider signal bandwidth, the stronger the ability of the array extension. At last, the algorithm was verified by a semi-physical simulation.Making an in-depth study on joint estimation of 2-D DOA for L-shape arrays, main researching how to utilize the structure of the array sensors and the data to improve performance. A fast algorithm based on the structure of the array sensors is proposed, which primarily through the use of pseudo-correlation matrix of the reduced-dimensional structure to achieve the purpose of quickly computing. the algorithm also has the automatic matching features. Also in L antenna array, a novel algorithm is proposed, which utilizes the physical structure and the relationship between signals to achieve automatic matching and array extension. This algorithm use Kronecker product and make use of some mathematical derivation of the matrix to conscious rearrange data. At last, an algorithm, which utilizes the array structure and the signal correlation time longer than the noise, is proposed. This algorithm can deal with colored noise and has the capability of array extension. Since the introduction of information on the physical structure of the antenna, the algorithm also has the ability to automatically match. In addition the accuracy of the three algorithms mentioned above performs well in similar algorithms.更多还原