基于子空间的超分辨测向算法及实现的研究

发布时间：2018-05-12 22:28

本文选题：电子战 + 超分辨测向　；参考：《江苏科技大学》2017年硕士论文

【摘要】：现代电子战离不开对目标的精确测向,随着电磁环境的日益复杂和电子侦察系统指标的不断提高,传统测向方法已经难以适应现代战争需求。超分辨测向算法以其优越的性能,已在一些实时性要求较低的系统中得到应用,随着软硬件实现能力的提高,使其在电子侦察领域的应用成为可能。本文以舰载复杂环境下电子战测向接收分机为应用背景,开展以多重信号分类(MUSIC)算法为代表的子空间类分解算法、低阵元数阵列测向性能,特定阵形下解相干算法以及算法实现等方面的研究工作,主要内容包括:1.分析MUSIC算法的基本原理与改进分类。简述了MUSIC算法适用的信号类型,分析了算法原理与特点,并给出了一种高效的噪声功率估计方法。2.细致比较了低阵元数天线阵形的测向性能。采用降维投影的方法,提出了均匀圆阵“无模糊”半径选取方法,利用阵列流形与子空间的相关系数简化了阵列误差的表示方法;从抗模糊性能、测向精度、各向等效性以及阵列误差敏感度等角度对奇偶阵元数均匀圆阵和三种一维线性阵列的测向性能进行了分析、对比与仿真。3.综合考虑成本与测向性能,研究本课题适用的解相干信号测向方法。分析了经典解相干算法的特性,将前后向空间平滑算法拓展到二维阵列,提出了中心对称平滑算法,该算法是一种能够提高阵元复用率的二维解相干算法;分析了不同角度入射相干信号对信号源数目估计的影响。4.拆分研究MUSIC算法的各模块的实现架构,以及方案的快速实现。分析并优化了现有奇异值分解算法的并行实现方案,给出了FPGA逻辑加速部分设计方案;为了均衡延迟与资源消耗,选用了三级细粒度的谱峰搜索方法;使用Xilinx SDSo C套件,在Z-Turn平台上完成算法的快速实现。
[Abstract]:Modern electronic warfare is inseparable from the accurate direction finding of targets. With the increasing complexity of electromagnetic environment and the continuous improvement of electronic reconnaissance system, the traditional direction finding methods have been difficult to meet the needs of modern warfare. Super-resolution direction finding algorithm has been applied in some systems with low real-time requirements due to its superior performance. With the improvement of the ability of software and hardware, it is possible to apply it in the field of electronic reconnaissance. In this paper, based on the application background of electronic warfare direction-finding extension in shipborne complex environment, the subspace class decomposition algorithm, which is represented by multiplex signal classification and MUSIC-based algorithm, is developed, and the direction finding performance of low array is obtained. The research work on decoherence algorithm and algorithm implementation under specific formation includes: 1. The basic principle and improved classification of MUSIC algorithm are analyzed. This paper briefly describes the signal types suitable for the MUSIC algorithm, analyzes the principle and characteristics of the algorithm, and presents an efficient noise power estimation method .2. The direction-finding performance of antenna array with low element number is compared in detail. By using the method of reducing dimension projection, the method of selecting the radius of uniform circular array "without fuzzy" is put forward, and the expression method of array error is simplified by using the correlation coefficient between array manifold and subspace. The performance of uniform circular array with odd and even array elements and three kinds of one-dimensional linear array are analyzed from the angles of equivalence and sensitivity of array error. Considering the cost and direction finding performance, this paper studies the direction finding method of decoherence signal. After analyzing the characteristics of classical de-coherent algorithm, the forward and backward spatial smoothing algorithm is extended to two-dimensional array, and a centrosymmetric smoothing algorithm is proposed, which is a two-dimensional decoherence algorithm which can improve the multiplexing rate of array elements. The influence of incident coherent signals at different angles on the number of signal sources is analyzed. This paper studies the implementation framework of each module of MUSIC algorithm and the fast implementation of the scheme. The parallel implementation scheme of the existing singular value decomposition algorithm is analyzed and optimized, and the design scheme of FPGA logic acceleration part is given. In order to balance delay and resource consumption, the three-level fine-grained spectral peak search method is selected, and the Xilinx SDSo C suite is used. Complete the fast implementation of the algorithm on the Z-Turn platform.
【学位授予单位】：江苏科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN97;TN911.7

【参考文献】