【作者】 文忠；

【导师】 张锡祥； 李立萍；

【作者基本信息】 电子科技大学 ， 信息获取与探测技术， 2007， 博士

【摘要】 随着雷达技术的广泛应用和发展，现代雷达的信号形式、传播环境已发生了很大的改变。现代雷达在时域、频域、空域的分布很广，非平稳、相干干扰、短时间序列也是常常面临的问题。研究能够适用于现代雷达信号环境且性能优良的多参量估计方法是信息探测领域的前沿课题之一。本文深入研究了处于现代电磁环境中的雷达信号时空多维参量估计算法。归纳起来，主要包括以下几个方面：1．研究了时频分析工具在非平稳雷达信号分析中的应用。提出了一种适合线性调频信号(LFM)的交叉项抑制时频分布。该分布是对Wigner-Ville分布的改进，通过分段补偿平滑的方式来抑制Wigner-Ville分布中产生的交叉项。2．研究了线性调频雷达信号的时频检测方法与基于解线调原理的特征参数估计方法。并针对超低信噪比的情况，提出一种非线性的弱LFM雷达信号的检测与参数估计方法。该方法利用Duffing振子构建广义滤波器组，并以其最大Lyapunov指数的符号作为系统状态的判断标准。根据信号特征，将LFM信号的检测问题转化为一个周期检测的问题，使其满足Duffing振子的检测条件。为提高信噪比，本文还提出分段相关平滑的方法，使检测和估计性能得到提高。3．分析了窄带阵列信号模型，研究了将时频分析与传统子空间测向方法结合的时频子空间测向算法。针对相干干扰问题，利用脉冲雷达的特点，提出一种新的估计相干信号源到达角(DOA)估计方法。本算法只需要K+1个阵元即可实现K个相干信号的DOA估计。而且，该算法对阵列形式约束较小。研究了短时间序列的波达方向估计问题，提出改进的伪协方差矩阵以及相应的DOA估计算法。4．分析了宽带阵列信号模型，并结合具体的非平稳雷达信号，推导了宽带线性调频信号的简化模型、完整模型以及宽带对称三角线性调频(STLFM)信号的阵列输出模型。提出了基于虚拟WVD的宽带LFM信号的到达角估计方法、基于模糊变换的宽带LFM信号的到达角估计方法以及基于数据分段虚拟阵元的宽带STLFM信号的到达角估计方法。仿真表明，提出算法比传统的TF-CSM方法精度更高，计算量更小。5．研究了极化敏感阵列多参量估计问题。分别提出演变周期谱与正交偶极子阵列结合算法，以及分数阶傅里叶变换与矢量阵元阵列结合算法，高精度地实现了LFM信号DOA与极化参数的联合估计。提出空间—极化分维方法，利用矢量天线阵元实现了相干信号的DOA与极化联合估计。研究了正交偶极子阵列的信息构成，提出基于该阵列的宽频段信号DOA与极化联合估计算法。为避免极子之间存在的互耦效应，提出一种由分离偶极子构成的阵列形式，实现了信号DOA与极化联合估计。6．研究了DOA与时延联合估计问题。针对雷达多目标回波信号，提出了一种新的高分辨、高精度的到达角和时延联合估计算法。更多还原

【Abstract】 The signal system and propagation envirment of the modern radar were changed greatly, along with the application and development of the radar techniques. Furthermore, the spans of the time, frequecy and space of the modern radar are very large, non-stationary and the coherent interference or short sequences appear frequently. The researches on multi-parameter estimation methods applied in the circumstance of the modern radar signals were advanced problem in the information domain. The estimation algorithms of spatio-temperal multi-parameters in the modern electromagnetic envirment were focused in this dissertation. The main contributions of this dissertation concentrate in a few aspects as follows:1. The time-frequency analysis tools applied in the nonstationary radar signals is discussed. And a new reduced crossterm distribution for linear frequency modulated(LFM) signals is presented. This approach is the modification of the Wigner-Ville distribution(WVD), which surppresses crossterm by segment compensation and smoothing.2. The time-frequency detection methods and character parameters estimation methods based on the dechirping are analysed. In order to deal with the LFM signals in the ultra-low signal-noise-ratio(SNR) case, a nonlinear algorithm for weak chirp signal detection and parameters estimation is presented. The generalized filter array is constructed by the Duffing oscillator, and Lyapunov exponential is exploited to judge the state of the system. According to the characteristic of the chirp signal, the problem of chirp signal detection is transformed to the period detection problem, which is satisfied the demand of Duffing oscillator system. Moreover, the segmented correlation smoothing method is proposed to enhance the SNR that improved the performance of the method.3. The narrow band model of the array output is analysed, and the conventional subspace methods combining the time-frequency analysis are introduced. To coherent interference, a novelty direction-of-arrival(DOA) estimation method is proposed exploiting the characteristic of pulse radar. K coherent signals can be processed with K+1 arrays sensors in this method. Moreover, the algorithm exhibits litte astrict to array geometries. The direction finding of the short sequences is investigated, and a modified pseudo-covariance and DOA estimation algorithm is presented.4. The wideband models of the array output were discussed, according to the alternative nonstationary radar signals, the simplified model and complete model of the wideband LFM signals are derived respectively, as well as wideband symmetric triangle LFM(STLFM) signals. And then, the DOA estimation methods for wideband LFM signals based virtual WVD, for wideband LFM signals based ambiguity transformation and for wideband STLFM signals based data segment virtual elements are proposed respectively. The simulations illustrate that the proposed approaches have higer precision and less computation than conventional TF-CSM methods.5. The multi-parameters estimation problems with polarization sensitivith anttenas array are studied. The algorithms of evolutionary period spectrum combing crossed dipole array and fractional Fourier transformation(FRFT) combing vector sensor array are presented respectively, which can implement the hgier accurate joint DOA and polarization estimations for LFM signals. The space-polarization freedom degree dividation was proposed, thus, the DOA and polarization of coherent signals are estimated with vector sensor array. Analysing the information strcture of the cross diploes, a DOA and polarization estimation method is presented for broadband signals. An new array with separated dipoles is proposed to obtain the joint DOA and polarization estimation for avoiding the mutual coupling of co-center poles.6. The problem of the joint DOA and time-delay estimation is discussed. For estimating the DOA and time-delary of the radar echo signals, a novelty high resolution and presicion algorithm is presented.更多还原