【作者】 王党卫；

【导师】 粟毅；

【作者基本信息】 国防科学技术大学 ， 信息与通信工程， 2006， 博士

【摘要】 超宽带雷达是一种新体制雷达,已成为现代雷达系统发展的重要方向之一。在这种新体制下,雷达获取目标和环境信息的能力在频域大幅度扩展,可同时激励目标瑞利区、谐振区及光学区散射特性,进而使提取目标和环境更为丰富的电磁波谱信息成为可能。UWB雷达目标电磁特征抽取与识别方法的研究对于探求雷达目标识别的新机理、新方法,提高精导武器和战场侦察系统目标识别能力具有重要理论意义和实用价值。本文系统研究了UWBR导体目标识别的理论和技术,主要工作包括:①开展了基于UWBR导体目标散射回波数学模型特征提取方法的研究。本文深入分析了导体目标散射回波后时、前时以及整体回波的数学模型,分别给出了导体目标散射回波后时的精确表示方法和散射回波的统一模型;接着,针对传统极点提取方法存在的前后时回波分割问题,通过对目标整体散射回波的ARMA模型逼近,结合SVD和TLS技术提出了一种基于ARMA模型目标极点提取算法,有效改善了极点提取的精度;随后,基于给出的散射回波统一模型开展了UWBR导体目标散射中心特性分析以及特征提取的研究,提出了一种GTD模型参数提取新方法,通过实验手段揭示了UWBR导体目标散射中心存在主散射分量这一物理现象,并分析了利用GTD模型提取导体目标主散射分量和散射回波建模的可行性。②针对基于模型特征抽取方法存在的不足,开展了基于UWBR导体目标散射回波稀疏特性的无模型约束特征抽取方法的研究。本文定义了信号的欠完备原子库稀疏表示问题,理论证明了欠完备原子库对信号空间的二元划分能力,给出了欠完备原子库的几点重要性质以及求解具有最小分解残差能量欠完备原子库的训练算法;随后,根据欠完备原子库的有关特性,提出了一种目标识别新特征——原子库特征、相应的抽取方法以及原子库特征最优原子库个数的搜索方法,并通过大量实验证明了该新特征的有效性。③利用导体目标极点特征开展了基于UWBR散射回波后时目标识别方法的研究。本文首先理论分析了导体目标回波后时的互相关特性,推导了回波后时互相关处理后信号交叉项不产生新极点分量的条件,给出了一种新的信号互相关函数估计方法,并针对Blanco等人提出的渐近无偏E脉冲目标识别方法存在的缺陷,结合新的互相关算法,提出了MAUAE目标识别方法;随后,针对低信噪比目标识别问题,理论分析了BE脉冲识别方案具有的优势及存在的问题,分别研究了BE脉冲方案的改进方法、实现技术以及快速目标识别方法,相继提出了MBE脉冲识别方法、E脉冲发射端合成技术以及BEGLRT快速目标识别方法,形成了基于极点特征的目标识别理论与系统实现框架,为基于目标回波后时的低信噪比目标识别提供了理论基础和工程化实现的初步方案。④针对基于回波后时目标识别存在的不足,分别使用提出的导体目标散射中心特征和原子库特征开展了基于目标整体散射回波目标识别方法的研究。基于似然比检测和匹配追踪理论,本文分别使用第二、三章提出的回波散射中心主散射分量GTD模型参数和原子库作为特征,提出了两种不同的目标回波重构方法以及相应的目标识别方案,并利用电磁计算和实测数据证实了这两种方法的有效性。更多还原

【Abstract】 Recently, the ultra wide-band radar, as a new system, has been an important development direction of the modern radar system. With the new system, radar target and environment scattering information will be excited more completely in frequency domain. The scattering characteristics in Rayleigh, resonance and optics domains can be measured together and more complete electromagnetic information of radar target and environment can be extracted. The theory investigation of feature extraction and target identification of ultra wide-band radar is of much theoretical and applicable significance in the exploitation of novel mechanisms and methods for target identification, as well as the improvement of the target identification performance in the precise guidance weapon and battlefield surveillance system.In this thesis the theory and technique of conducting target identification based on the new radar system are investigated, and main works are accomplished as follows:Firstly, the feature extraction methods based on the mathematics models of the conducting target’s ultra wide-band scattering echo are investigated. The mathematics models of the whole scattering echo, late-time and early-time components are analyzed respectively in this thesis. A uniform scattering model and exact expression are given respectively for the whole scattering echo and its late-time component. Then, in view of the difficult partition to the late- and early-time component of the scattering echo, through an ARMA representation to approximate the whole target scattering echo a pole feature extraction method is proposed with SVD and TLS (Total least square) techniques, which improves the pole extraction precision. Subsequently, based on the uniform scattering model, the characteristic analysis of UWBR conducting target scattering center and feature extraction are investigated respectively. An extraction method of GTD model parameter is proposed, and then based on the proposed method a physical phenomenon that there exists a main scattering component among complicated scattering behaviors of the scattering center of conducting target illuminated by ultra wide-band radar is indicated in the experimental manner. Furthermore, the feasibilities of GTD model for main scattering component extracting and the whole scattering echo modeling are analyzed respectively.Secondly, aiming at the deficiencies in feature extraction based on the model, the construction and extraction techniques based on the sparse property of the whole scattering echo are investigated for the feature without the model restriction. The signal sparse representation with the undercomplete atom dictionary is defined, the binary partition capability of the undercomplete atom dictionary to the signal space is theoretically proved and several important properties of the undercomplete atom dictionary are given, as well as a training algorithm for the undercomplete atom dictionary with least decomposition residual energy. Furthermore, a novel feature for target identification, named as atom dictionary feature is proposed and validated, including an extraction algorithm for the atom dictionary feature and a search algorithm for atom number optimization of the atom dictionary feature.Thirdly, pole feature-based target identification using late-time component of the UWBR conducting target scattering echo is mainly investigated. The cross correlation characteristic of the target late-time component is theoretically analyzed, the condition not producing the new pole component is concluded, and then a new cross-correlation estimator is proposed. According to the new cross-correlation estimator, aiming at the deficiencies in the asymptotically unbiased E-pulse target identification method proposed by Blanco, a method named as MAUAE is proposed. Subsequently, with a view to the deficiencies in the MAUAE method, the superiority and shortcoming of the BE-pulse scheme are theoretically analyzed. The modified BE-pulse identification scheme, engineering realization technique of the E-pulse waveform, and the fast target identification method named as BEGLRT are proposed in succession. As a result, the pole feature-based theory and system sechmes for target identification are established, which provide the pole feature-based target identification in low SNR with the theory basis and elementary engineering realization scheme.Finally, in view of the deficiencies in target identification based on late-time component, the whole scattering echo-based target identification is investigated with the proposed scattering center and atom dictionary features for ultra wide-band radar. Using the likelihood ratio test and matching pursuit technique, the echo reconstruction and target identification methods with the GTD model parameters of the main scattering component of the scattering center and atom dictionary from chapter 2 and 3 as features are proposed in this thesis respectively. Furthermore, based on the calculated and measured transient response by author, the proposed echo reconstruction and target identification methods are validated.更多还原