【作者】 石志广；

【导师】 郭桂蓉； 赵宏钟；

【作者基本信息】 国防科学技术大学 ， 信息与通信工程， 2007， 博士

【摘要】 现代战场环境复杂多变以及信息化战争对雷达信息处理能力要求的提高使得杂波下的信息处理问题尤为关键,而高分辨雷达技术的发展改变了雷达杂波特性,给杂波下的信息处理技术提出了新的挑战。本文结合背景需求,针对高分辨雷达杂波特性分析与建模以及杂波下的信号处理方法两大问题展开研究,主要内容包括以下四个方面:第二章基于统计和随机过程理论对实测海杂波数据进行特性分析。首先总结了杂波统计特性分析的基本理论和方法,重点分析比较了K分布杂波模型不同参数估计方法的性能,提出了基于粒子滤波器的K分布参数估计新方法;其次,结合项目背景,对实测海杂波数据进行分布拟合检验、时空相关特性以及谱特性分析,并基于SIRP/ZMNL实现了时空二维相关K分布杂波的仿真。第三章基于复杂性理论对高分辨海杂波数据的非线性和复杂性进行研究。首先简单回顾复杂性理论的发展历史与主要内容,定性分析了海杂波的复杂性特点;其次运用相空间重构技术、替代数据方法、递归图方法以及复杂性测度理论等对海杂波时间序列的非线性、可预测性、复杂性进行定量分析;再次,对海杂波的多重分形特性进行实验判定,分析了海杂波多重分形特性的的变化规律,提出了杂波的小波多重分形模型及其仿真方法;最后利用一类非线性时频分析方法-希尔波特-黄(HHT)变换对杂波数据进行分析,提出了基于HHT能量分布特征的杂波下目标鉴别新方法。第四章研究非高斯杂波下的小目标检测问题,主要探讨了几类新的检测方法。①基于RBF神经网络非线性预测的检测方法,提出先通过小波分析降噪,提高数据的可预测性,再用目标单元数据进行神经网络训练的方法;②K分布杂波下基于随机共振的小目标检测方法,对K分布杂波下的最优检测处理器进行了推导,并与基于阈值处理的随机共振系统进行对比,提出基于阈值随机共振的次优检测方法;③提出了分数阶循环平稳信号的概念,将其应用于Alpha-Stable分布杂波下LFM信号的检测与估计。第五章研究杂波下一维散射中心参数估计以及目标的运动参数估计问题。首先研究了强杂波背景下GTD散射中心模型的定阶与参数估计问题,提出基于贝叶斯原理和RJ-MCMC算法的GTD模型联合参数估计与定阶方法;其次,针对非高斯杂波中尖峰数据的存在,采用基于M估计的稳健算法进行散射中心参数估计,并利用协同PSO算法解决M估计中的优化问题;最后研究了非高斯杂波下目标运动参数估计问题,提出一种可应用于“静态参数”序贯估计的粒子滤波器改进算法。最后对本文内容进行了系统的总结,对论文中需要进一步研究的工作进行了探讨。更多还原

【Abstract】 The radar signal processing in clutter becomes a vital problem in the information based modern war because of the complex and variable battlefield and the increasing demand on the information processing ability of radar.Moreover,the characteristic of radar clutter has changed as the resolution of radar increases,and this puts forward a new challenge for modern radar signal processing methods.Evoked by the practical demand of several projects, we carry out research on the clutter characteristic of high range resolution radar and the signal processing method under clutter.The work contains four parts:The second chapter analyzes the statistical characteristic of some measured sea clutter data.Firstly,the fundamental theory and method for analyzing the statistical characteristic of radar clutter is introduced.Different methods for estimating the model parameters of K-distributed clutter are compared and a new estimation method based on particle filter is proposed.Secondly,the statistical characteristic of the measured sea clutter data in two special projects is analyzed,including distribution fitting,time-space correlation analysis and spectrum analysis.The simulation method of spatial-temporal correlated clutter is realised based on SIRP and ZMNL.The third chapter studies the complex characteristic of high range resolution sea clutter based on the complexity theory.Firstly,the evolution and principle of the complexity theory is briefly introduced,and the complexity of sea clutter is analyzed.Secondly,the nonlinearity of the sea clutter in time domain is testified by phase space reconstruction and surrogate data method;the predictability and deterministic of high resolution sea clutter is analyzed using recurrence plot;the complexity of clutter and target echo is analyzed quantificationally using complexity measuring theory.Thirdly,the multi-fractal of sea clutter is tested experimentally; the varying property of the multi-fractal is discussed;and the simulation method of sea clutter based on wavelet-multi-fractal model is proposed.Lastly,we use a new time-frequency analysis method,i.e.,the Hilbert-Huang Transform(HHT),to compare the decomposed characteristic of clutter and target echo.A new way for target-clutter discrimination is proposed based on the different HHT energy characteristic of the target and the clutter.Chapter 4 focuses on the detection of small targets in non-Gaussian clutter.Several new detection methods are studied:①Detection method based on RBF neural network.We propose an improved algorithm which has a denoising step using wavelet decomposition and trains the network by target bins.②Small target detection method in K-distributed clutter based on stochastic resonance.The optimal detector under K-distributed clutter is deduced.A sub-optimal detector based on the quantizers is also proposed by comparing the optimal detector with the multilevel quantizers.③A detector based on the proposed principle of fractional lower order cyclostationary sigal is used to detect and estimate the LFM signals in alpha-stable clutter.Chapter 5 studies the parameter estimation problem in clutter,i.e.,estimating scattering center parameters from high resolution range profiles and estimating the motion parameter of targets in clutter.Firstly,the joint model selection and parameter estimation problem of GTD model under heavy noise is studied and a new method based on Bayes principle and RJ-MCMC algorithm is proposed.Secondly,a robust M-estimator is proposed to suppress the spiky data in non-Gaussian clutter when estimating scattering center parameters and the CPSO(co-operative particle swarm optimization) algorithm is used in the optimization process of the M-estimator.Lastly,the motion parameter estimation problem of targets in non-Gaussian clutter is studied and an improved algorithm based on the particle filter for the sequential estimation of the "static parameters" is proposed.A systematic conclusion together with some further discussion on future work is given at the end of this dissertation.更多还原