一种探地雷达目标参数估计方法的研究

在探地雷达应用中,确定地下目标的介质参数具有重要意义.为了对探地雷达地下目标参数进行估计,提出了将探地雷达偏移处理和时域目标参数反演结合起来的方法,该算法首先通过偏移处理,确定目标的位置和分布范围,然后采用FDTD和随机逼近优化法进行迭代运算估计目标参数.与传统基于频域的方法相比,该方法具有易于实现、处理速度快等优点.并且通过对模拟数据的处理,验证了该方法的可行性.     

矩形空洞探地雷达成像曲线厚度变化规律研究

由于城市基础建设不断地翻新改造,引发城市地下空洞滋生问题。通过探地雷达扫描技术,对地下矩形空洞整个探测过程进行研究,将探测过程划分为四个阶段,研究了成像曲线厚度变化规律。对探地雷达水平运行时矩形空洞的成像曲线厚度变化规律进行分析,对不同大小下的矩形空洞点测时曲线厚度进行研究,并通过实际探测案例来解释矩形空洞成像曲线厚度变化规律,结果表明探地雷达探测矩形空洞时其成像曲线厚度为先增加后减小的对称曲线,其曲线厚度最大处在图像上为水平弧段两端,位于实际矩形空洞对角线附近;雷达点测时成像曲线厚度随矩形空洞变大而增加,在图像上表现为空洞越大线条越粗。     

多相离散随机介质探地雷达三维数值模拟

沥青混凝土是典型的多相离散随机介质, 为了研究探地雷达波在这种介质中的传播特征, 基于随机介质模型理论与统计学方法, 研究了沥青混凝土芯样的空间随机分布统计特征、各组成物质体积百分比、自相关函数及其特征参数(如自相关长度、自相关角度等), 基于量化约束多相离散随机介质模型建模算法, 构建了与之对应的多相离散随机介质三维模型.采用时域有限差分法建立了商用探地雷达蝶形天线的三维模型, 开展了多相离散随机介质探地雷达三维数值模拟.研究结果表明:多相离散随机介质模型不仅能全面、准确地描述沥青混凝土的介质特征, 而且可用于描述其他的复杂非均质混合物, 其模拟剖面与实测剖面相符, 有利于指导探地雷达沥青混凝土质量无损检测实测剖面的数据解译.     

基于多测量向量模型的极化探地雷达成像算法

针对极化探地雷达( GPR)工作过程中目标成像空间的联合稀疏性,提出了一种基于多测量向量模型的极化探地雷达成像算法。在建立极化探地雷达回波信号模型的基础上,利用各极化通道测量数据的联合稀疏性将各个极化通道的测量数据等效成多测量向量( MMV ),通过多任务贝叶斯压缩感知( MT-BCS)算法对各个极化通道的测量数据进行联合处理从而实现各个极化通道对应的探测场景反射率的重建。基于时域有限差分( FDTD)法的仿真数据处理结果表明所提成像算法在目标位置重建的准确性和背景杂波抑制能力上均优于单测量向量( SMV)模型的极化探地雷达成像算法。     

基于遗传算法的探地雷达层状介质参数反演算法

探地雷达因其无损性与高效性,逐渐成为道路检测、地下勘探等领域的有效探测手段,而探地雷达的传统厚度反演算法存在无法有效检测分层介质厚度、相对介电常数等参数的问题,具有较大的多层介质参数反演误差。因此,本文提出了一种基于遗传算法的层状介质参数反演算法。算法从时域角度出发,结合共中心点方法,设计了基于遗传算法的优化模型,反演地下层状介质结构的参数信息。通过实验仿真,本文所提方法可以应用于地下多层介质,较为准确的反演出层状介质参数信息。      

基于NUFFT的机载探地雷达后向投影成像算法

由于后向投影算法可以精确补偿电磁波在介质表面发生的折射效应,因此它在机载探地雷达成像技术领域具有较强的工程实用价值。但传统后向投影成像算法存在计算量大难以实时实现的问题,针对上述问题,文中提出一种基于非均匀快速傅里叶变换(NUFFT)技术的机载探地雷达快速后向投影成像算法。通过对基于时域有限差分法产生的仿真数据进行处理,验证了所提成像算法的有效性和快速运算能力。     

用于探地雷达系统仿真的行波天线等效模型及应用

为了能够建立一种简单、通用且不需对实际天线建立物理模型的等效模型,提出了一种用于冲激脉冲探地雷达系统时域仿真的行波天线等效模型,该模型仅需考虑天线上的电流分布,将时域有限差分算法嵌入该模型,实现天线自由空间辐射场的计算,并将计算结果与加载领结天线辐射波形进行对比,通过对雷达半空间扫描进行仿真,将扫描结果与探地雷达实测A扫,B扫数据进行对比,可以实现较好的吻合,验证了模型的正确性,为模型应用于2种极化方式考察和多天线配置方案考察中奠定技术基础。     

一种用于准单站配置的步进频探地雷达建模方法

探地雷达是一种被广泛使用的无损检测技术。利用探地雷达对分层媒质进行全波反演时,构建精确的探地雷达正演模型具有十分重要的意义。该文提出一种可用于准单站配置的步进频探地雷达的建模方法。在该模型中,探地雷达系统及其与分层媒质间的相互作用被表示成线性方程,天线对雷达信号的影响被表示为只与频率有关的传输函数。为验证模型准确性,该文在实验室条件下搭建了准单站配置的步进频探地雷达系统,并对已知厚度的石膏板和木板的雷达测量信号进行全波反演。反演结果表明:石膏板和木板的厚度估计误差均不超过0.3 mm,验证了所提出的正演模型具有高准确度。利用石膏板和木板搭建分层模型,该文进一步比较了准单站配置和单站配置步进频探地雷达系统对介电常数差异较小的分层媒质的反演性能。实验结果表明:准单站配置探地雷达能获得更精确的反演参数。通过对分界面反射信号的信噪比估计可知,准单站配置比单站配置探地雷达系统能获得高出约10 dB的信噪比,因此具有更好的反演性能。      

一种反演地下介质参数的新算法

在探地雷达实际工程应用中,介质参数(主要是介电常数)是使用者正确定位地下目标深度或介质层厚度所必需的数据。该文提出一种基于MUSIC谱估计的地下介质参数反演算法,该算法通过对LFMCW GPR回波信号的时延和幅度估计,计算出精确的介质参数并进一步得到正确的地下介质结构。仿真结果表明:该文提出的反演算法在反演精确度上远远优于传统的电磁反演算法。     

脉冲探地雷达的模拟计算

本文在给出Debye型色散媒质中2．5维时域有限差分法（2．5D－FDTD法）迭代公式的基础上，对无载频脉冲波在不同色散媒质中的传播特性进行了计算，分析了脉冲产生畸变的原因，并提出对部分畸变脉冲进行整形的方法。分别对地下单体目标和群体目标的雷达回波电平图进行了模拟计算，并与实际无载频脉冲探地雷达的探测结果进行比较，二者有较好的一致性，证实了本文所给计算公式的正确性。另外，还分析了土壤参数对雷达探测深度和分辨率的影响。     

基于CFS-PML的高阶有限元在地质雷达数值模拟中的应用研究

In the numerical simulation of ground penetrating radar (GPR), to attenuate the evanescent wave generated by the thickness of the truncated area or the influence of parameters, a new method is proposed to use the complex frequency shifted perfectly matched layer (CFS-PML) as the absorption boundary and combined with the higher-order finite element method (HO-FEM) in the frequency domain to solve the distribution of the total electromagnetic field components. The method improves the properties of the dielectric intrinsic tensor matrix in the truncated domain and uses the matrix frequency shift parameter α to absorb the evanescent wave and improve the absorption performance at the grazing angle. Numerical experiments are conducted on the method. Compared with the traditional PML, the accuracy of the HO-FEM method combined with CFS-PML is significantly improved, the reflection error is reduced by 10?30 dB extra, and the calculation time is saved by 17%?30%. In addition, the method can be applied to the calculation of the total electromagnetic field of a complex geological structure model, which provides a new method for the numerical simulation of geological lidars.     

基于波场互相关的探地雷达快速自聚焦成像

传统探地雷达偏移成像算法计算复杂度高、准确性低,且需要准确已知介质的电性参数。针对这些问题,本文提出了一种基于波场互相关的探地雷达快速自聚焦成像算法,利用水平分层介质频域格林函数的平移不变特性和快速傅里叶变换减小了对计算和存储资源的需求,并能够实时对地下目标进行高分辨成像;通过引入时间相位因子得到不同聚焦时刻的偏移图像,然后基于图像熵最小准则获取最优成像结果,解决了由于介质参数未知而导致的图像散焦问题。仿真结果验证了所提算法在介质参数未知情况下具有良好的自聚焦性能。      

Radar sensing of thin surface layers and near-surface buriedobjects

A robust ground penetrating radar (GPR) signal processing approach is developed and applied to the sensing of surface soil and ice/water layers as well as near-surface buried objects. The principal technique relies on a reference set of waveforms, which are tested for optimal matching with measured radar reflections to be analyzed. In principle, the reference set can be based on measurements as well as on model output, with the latter employed. Even when layers are quite thin relative to the incident wavelength or pulse, the approach provides accurate information on layer characteristics, particularly thickness. The method assumes a well-defined transmit signal and hence is best used with antennas elevated shove the surface, In tests using UHF pulses on lab and field ice and on thawing soil, the system offers approximately an order of magnitude improvement in layer resolution relative to more traditional methods. Also, in the process of its signal-matching calculations, the procedures provide a numerical indication of the reliability of each result. In application to thawing soil, simulations suggest that one can address a wide variety of conditions using quite a limited set of reference signals. By detecting the thin-layer effect from soil over a buried metallic object, the system also locates near-surface targets when their reflections cannot be separated in time from ground surface returns. An alternative system also succeeds in detecting near-surface objects under the same conditions by detecting wavelet dispersion. This is done without reliance on specific details of transmit wavelet or reflected signals. A mine-like target is detectable in a wet clay soil when the ground is frozen but not when it is thawed     

Electromagnetic Subsurace Radar Using the Transient Field Radiated by a Wire Antenna

A radar based on the propagation of a short pulse of current along a wire parallel to the ground surface is described. The wire acts both as a traveling-wave transmitting antenna and as a receiving antenna. For applications to civil engineering, the depth of exploration may be less than 1 m, the purpose being, for example, to detect discontinuities situated near the ground surface, to measure the thickness of a thin layer (10-50 cm), or to characterize a change in its reflection coefficient. Therefore, the width of the pulse of current is about 1 ns. The theoretical approach shows that the wire must be situated at about 10 cm above ground and that there is an optimum position for the current probe that detects the induced current due to a reflecting layer. Several experiments will also be described showing the feasibility of the method.     

宽带天线罩和雷达吸波材料的优化计算

利用Dixon优化算法优化各介质层的厚度,并采用多层结构,使在确定的频率点上天线罩对电磁波有很好的透射特性,雷达吸波材料有较强的吸收特性;而且达到宽频带的要求。计算结果表明,优化的天线罩对固定频率的电磁波的透射系数的幅值接近最大值1,而附在金属表面的雷达吸波材料对固定频率的电磁波的反射系数的幅值接近最小值0;并且随着介质层数的增加,频带明显展宽。     

基于二维时频特征的地雷目标检测

利用车载前视超宽带地表穿透雷达进行大区域地雷探测是探雷的发展方向,目前制约车载前视超宽带地雷穿透雷达探雷实用化的关键问题是难以提取一致性较好的地雷目标特征。文中以地雷目标的斜距 方位 频率 反射角四维散射函数为基础,提取二维时频域特征图像。然后对时频图像进行特征选择,并实现目标和杂波的分类。为降低分类器训练误差以提高其泛化性能,将特征选择和分类器设计有机结合。同时为保证地雷探测过程中具有恒定的探测率,在分类器训练过程中以恒检测率下的虚警率作为代价函数,通过实测数据验证,该方法适用于车载前视超宽带地表穿透雷达,最终得到的分类器能提供恒定的检测率,并且具有很好的泛化性能。      

A signal-processing algorithm for the extraction of thinfreshwater-ice thickness from short pulse radar data

A radar data processing algorithm is presented which can estimate freshwater ice thicknesses both greater than and less than the normal minimum resolvable thickness. The basis for the algorithm is a model of the geophysical system and the resulting phase signature of the transfer function. The algorithm attempts to isolate this signature and estimate the relevant parameters. In principle, the method can be used to detect layering in other dielectric media. The algorithm was based on a simple model for scattering from a thin-ice layer and refined using radar returns taken over a range of thin-ice thicknesses in a laboratory environment. The method was also applied to data obtained with two different radar frequencies and over a wide range of thickness and surface roughness. The results were compared to thickness estimates from classical deconvolution methods     

Reconstruction of multilayered lossy dielectrics from plane waveimpulse responses at two angles of incidence

Motivated by the radioglaciology inverse problem, the authors present new algorithms for reconstructing a lossy, stratified dielectric from its impulsive plane wave reflection responses at two different angles of incidence. Novel features of these algorithms include: 1) a digital signal processing formulation that does not require discretization of continuous equations; 2) use of the asymmetric Levinson algorithm for rapid solution of the forward and inverse problems; 3) a novel use of layer stripping ideas, featuring iteration between the forward and inverse problems, with each iteration recursively determining another layer of the medium; and 4) another recursive algorithm for determining the bottom lossy half-space from reflection data only. Numerical examples illustrate the new algorithms on the reconstruction of a synthetic but realistic layered ice shelf from both noiseless and noisy radar reflection data