【作者】 秦洪峰；

【导师】 黄建国；

【作者基本信息】 西北工业大学 ， 信号与信息处理， 2002， 博士

【摘要】 多目标定位技术在声纳、雷达、通信、海洋开发、地质勘探、国防建设和生物医学工程等众多应用领域中占有十分重要的地位。论文从实际应用的角度出发，全面论述了多目标多参数估计理论和方法，并紧密结合水下目标探测与跟踪系统的工作特点，从理论分析、计算机仿真和水池实验三方面对水下多目标定位关键技术进行了系统、深入的研究。主要研究工作概括如下： 1．深入分析了波束域高分辨方位估计性能，捉出了一种基于均匀线列阵的密集波束域MUSIC(DBMUSIC)法。研究表明，在低信噪比的条件下，DBMUSIC法的高分辨性能明显高于原波束域MUSIC(BMUSIC)法。 2．系统研究了阵元域和波束域高分辨方位估计方法的稳健性，建立了阵列误差模型，针对均匀线列阵提出了一种基于阵列输出协方差矩阵的加权Toeplitz化高分辨方位估计方法，能够显著提高存在阵列误差情况下方位估计方法的高分辨性能。给出了一种简单有效的加权系数表达式，并分析了加权系数对方位估计性能的影响程度。同时结合不同的高分辨方法相应地提出了加权Toeplitz化MUSIC(TMUSIC)法、加权Toeplitz化Mini-Norm(TMini-Norm)法、加权Toeplitz化波束域MUSIC(TBMUSIC)法以及加权Toeplitz化密集波束域MUSIC(TDBMUSIC)法。研究表明，存在阵列误差条件下，加权Toeplitz化高分辨方位估计方法具有稳健性高、估计性能良好、易于工程实现等优点。 3．根据多目标定位的技术特点，提出了一种方位-时延参数联合估计ESPRIT-TDF法。该方法利用高分辨旋转不变技术和高分辨时延频率技术，可在单次目标回波内实现多目标信号的方位和时延参数的高分辨联合估计，具有运算量小、分辨能力强、估计精度高等特点，能够对多目标进行高分辨精确定位。同时还利用减少交叉项分布(RID)实现水下多目标信号的时延-频率联合估计。研究表明，RID具有良好的多目标分辨能力和时频参数估计精度。 4．通过系统的实验研究，进一步验证了论文所提出的多目标定位方法的正确性和有效性。结合水下多目标定位的实际应用，利用多目标参数估计与定位水池实验系统对多目标方位估计方法、方位-时延联合估计方法以及时频参数联合估计方法进行了全面深入的水池实验研究。针对不同的目标情况，在方位估计方面分别研究波束形成法、MUSIC法、TMUSIC法、M ini一Norm法、TMini一Norm法、BMUSIC法、TBMUSIC法、DBMUSIC法、TDBMUSIC法的估计性能，在方位一时延联合估计方面分析了ESPRIT一TDF法的估计性能，在时频参数联合估计方面讨论了RID的估计性能。实验结果对这些多目标定位方法的实际应用具有重要作用。 5.首次研制了多目标定位分析软件系统。该软件系统共有六十多种目标参数估计方法，包括多目标方位、距离、径向速度的常规估计方法、高分辨估计方法以及论文中所提出的多目标定位方法，具有功能强大、操作简单、容错性强等特点，既能够对各种估计方法进行仿真研究，还可以分析各种实际数据，并能够直观、全面、有效地显示参数估计及目标定位结果。更多还原

【Abstract】 Multiple source localization technique plays a very important role in sonar, radar, communication, marine resources exploitation, geologic exploration, national defense, biomedical engineering, etc. Aiming at the applications in engineering, the principles of parameter estimation are discussed comprehensively in this dissertation. Considering the characteristics of underwater detecting and tracking system, the key techniques of underwater multiple source localization are systematically studied by theoretical analysis, computer simulation and experiment in the water tank. The main results and achievements are summarized as follows:1. The performance of beamspace high-resolution direction-of-arrival (DOA) estimation is analyzed. A dense beamspace MUSIC (DBMUSIC) algorithm for the uniform linear array (ULA) is proposed. It is shown that the performance of DBMUSIC is superior to that of the original beamspace MUSIC (BMUSIC) in the low signal-to-noise ratio (SNR) case.2. The robustness of high-resolution DOA estimators in the elementspace and beamspace is investigated thoroughly. An error model of array is established. A novel class of weighted Toeplitz high-resolution DOA estimators based on the covariance matrix of ULA output is proposed, which can obviously improve the performance of high resolution in the presence of array errors. A simple and effective expression of weight coefficient is given. Moreover, the influence of weight coefficient to the performance of DOA estimators is discussed. Referring to different high-resolution approaches, the corresponding weighted Toeplitz MUSIC (TMUSIC) algorithm, weighted Toeplitz Mini-Norm (TMini-Norm) algorithm, weighted Toeplitz BMUSIC (TBMUSIC) algorithm and weighted Toeplitz DBMUSIC (TDBMUSIC) algorithm are presented respectively. It is demonstrated that the weighted Toeplitz high-resolution DOA estimators in the presence of array errors possess the properties of high robustness, well performance, and is easy to be implemented.3. Considering particular characteristics of multiple source localization, a novel and efficient ESPRIT-TDF algorithm of joint estimation of DOA andtime-delay is proposed. Within one single source echo, both DOA and time-delay can be estimated simultaneously by this algorithm via high-resolution rotational invariance technique and high-resolution time-delay frequency (TDF) technique. The ESPRIT-TDF algorithm has features of low computation, good precision, and high resolution both in DOA and time-delay. It is suitable for multiple source exact localization. Meanwhile, the performance of joint estimation of time-delay and frequency using the reduced interference distribution (RID) is analyzed comprehensively. It is demonstrated that RID has good performance in resolution and precision.4. Through experimental study, the correctness and efficiency of proposed algorithms are proved. A series of experiments by using the experimental system in the water tank for parameter estimation and source localization are conducted on DOA estimation, joint DOA and time-delay estimation, and joint time-frequency estimation. In the case of different parameters, the performance of the estimation algorithms is investigated, including beamforming, MUSIC, TMUSIC, Mini-Norm, TMini-Norm, BMUSIC, TBMUSIC, DBMUSIC, TDBMUSIC, ESPRIT-TDF, and RID. Experimental results are shown that the algorithms above are significant in the engineering applications.5. An analysis software system of multiple source localization is developed for the first time. This software system consists of more than 60 estimation algorithms for the source parameters (DOA, range and velocity), including conventional algorithms, high-resolution algorithms, and the proposed novel algorithms. It possesses several advantages, such as powerful processing and output functions, easy operation, and high error-tolerance. This software system can not only for investigating the performance of paramet更多还原