【作者】 杜金香；

【导师】 李志舜；

【作者基本信息】 西北工业大学 ， 信号与信息处理， 2004， 硕士

【摘要】 高分辨方位估计是现代信号处理的重要内容之一，随着微电子集成及数字技术的发展，其在声纳、雷达等许多领域有着很好的应用前景。高分辨算法最初都是在窄带信号的假设条件下提出的，由于宽带信号含有更丰富的信息，有利于目标检测、参量估计和目标特征提取，因此，对于宽带信号处理的研究越来越多。恒定束宽波束形成和宽带源方位估计是宽带阵列信号处理的重要研究内容。论文从实际工程应用的角度出发，系统研究了恒定束宽波束形成和宽带高分辨阵列信号处理的相关理论和实现技术，主要研究内容包括： 1．系统概要地论述了阵列信号处理的基本概念和理论，给出了宽带阵列信号处理的模型。 2．研究了宽带恒定束宽波束形成器的两种设计思想，并重点分析了计算阵元权系数的最小二乘法和空间重采样法，比较了二者的性能，分析了恒定束宽波束形成器的频域和时域实现原理。 3．研究了宽带高分辨阵列信号处理的典型算法——相干信号子空间算法(CSSM：Coherent Signal Subspace Method)。分析了CSSM方法分辨相干源的实质，即通过频域平均消除了相关函数矩阵的奇异性；分析比较了几种不同聚焦矩阵对算法性能的影响。 4．研究了宽带信号波束域处理技术。分析比较了波束域处理与阵元域处理的分辨能力以及估计性能，并从降低分辨门限和估计方差角度考虑，给出了转换矩阵的构造方法。 对于转换矩阵(恒定束宽波束形成矩阵)正交化过程对恒定束宽性能的损失提出了一种补偿措施，即在正交化过程中对不同频率对应的波束进行加权，称之为加权正交化恒定束宽波束形成矩阵。并且仿真实验验证了加权正交化恒定束宽波束形成矩阵既具有良好的恒定束宽性能，又具有与一般正交化恒定束宽波束形成矩阵相近的分辨性能和估计性能。 5．完成了恒定束宽波束形成以及宽带源方位估计算法在DSP硬件系统中的实现。分析了系统的硬件结构，介绍了ADSP21160的结构特点和软件设计方法。采用模块化设计方法，完成了系统软件的设计、调试。针对系统对于算法实时性的要求，对于计算量集中的子模块设计了合理的实现方案。进行了基于DSP硬件系统的仿真实验和实验室实验，实现了3路恒定束宽波束形成和宽带相干源波束域DOA估计的实时处理。更多还原

【Abstract】 As an important part of the modern signal processing, high-resolution direction of arrival estimation technique has a promising perspective in many fields such as radar and sonar, with the development of the microelectronic and digital techniques. The analysis of wideband signals will improve the resolution capability and accuracy of estimation, for more frequency components can be synthetically used. Constant beam-width beam-former and direction of arrival of sources for wideband signals are the important parts of wideband array signal processing. Aiming at the practical applications on engineering, the dissertation studies the theories and application technologies of constant beam-width beam-former and DOA estimation for wideband signals. The main results and achievements are summarized as follows:1. The fundamental concepts and theories of array signal processing are discussed summarily, and the model of wideband array signal processing is set up.2. The two thoughts of designing constant beam-width beam-former are studied, and the least square method and spatial resample method to calculate the weighting coefficients of the array elements are analyzed and compared, and the constant beam-width beam-former is designed, both in frequency and time domains.3. The typical wideband high-resolution method-the coherent signal subspace method (CSSM) is studied. We analyzed its nature of resolving coherent sources and compared the influence of different focusing matrices. Simulations are conducted to test the performance of these focusing matrices.4. The wideband beam-space high-resolution technique is studied. The differences on resolution capability and the performance of estimation between beam-space processing and element-space processing are analyzed. Transforming matrices are designed for the consideration of depressing the resolution threshold and estimation variance.A method is proposed to make for the loss of the orthogonal converting matrix’s performance on constant beam-width, that is giving different frequency components different weighing values during orthogonal process, which is called weighted orthogonal converting matrix. Simulation results show that the weighted orthogonal converting matrix not only has good performance on constantbeam-width, but also has the same performance on resolution and estimation as those of common orthogonal converting matrix.5. The constant beam-width beam-former and wideband DOA estimation algorithms are realized in the DSP hardware systems. The system hardware architecture is analyzed, and the properties of ADSP2116x and the methodology of software development are presented. The system software is developed and debugged in modularization. Reasonable designing scenarios are proposed for the key modules, for the consideration of algorithms’ being performed in real time. The simulation results based on the target board show that the system running properly.更多还原