裂隙岩含水层污染探测多分集阵列探地雷达技术

Fractured-rock aquifers are widely distributed near land surface and are highly susceptible to contamination from human activities. The research on ground-penetration radar (GPR）technologies for characterization and monitoring of contamination in fractured-rock aquifers, is of great significance in guaranteeing China's groundwater safety and promoting the nation's green ecology development. Because of the complex heterogeneity of fractured rock aquifers and the deficiency in data acquisition and processing methods, the existed GPR technologies are not able to cost-effectively infer fracture and fracture-zone locations, orientations and properties, and it is still difficult to develop informed interpretation of geophysical results. With specific applications oriented like characterization and monitoring of contamination in fractured-rock aquifers, this project is based on the existed research of our project group in hidden heterogeneity target detection and imaging. Closely following the development trend of high accurate GPR and intelligent GPR, and to counteract non- linearity and ill-posedness issues of inverse scattering problems in wideband GPR subsurface imaging, our project group aims at further researches on multiple diversity array GPR detection and imaging methods appropriate for characterization and monitoring of fractures and contamination in fractured-rock aquifers. By joint optimized design of transmit waveforms in multiple diversity array and establishing tensorization analysis models of the received returns, the corresponding detection and imaging methods are investigated with the thought of cognitive radar and deep learning. multiple diversity array GPR detection and imaging methods appropriate for characterization and monitoring of fractures and contamination in fractured-rock aquifers can then be systematically proposed. This project is expected to obtain original findings in qualitative or quantitative interpretation of GPR geophysical results and relevant signal and information processing methods. The research of this project can prospectively provide the key fundamentals to design the advanced GPR equipments of characterization and monitoring of contamination in fractured-rock aquifers.

裂隙岩含水层广泛分布在地表附近,极易受到人类活动的污染，开展GPR监测技术研究, 对保障我国地下水安全、促进绿色生态发展具有十分重要意义.由于地下裂隙岩含水层复杂的非均匀性及信息获取和处理方法的局限性,现有GPR技术无法有效推断裂隙的位置、走向及特性,在GPR结果智能解译方面也存在困难.本项目组拟在长期开展地下隐藏异质体探测和成像研究的基础上, 面向GPR在地下水污染探测等具体应用,瞄准未来GPR高精度、智能化发展需求,克服宽带GPR表层逆散射成像的非线性和不确定性问题,进一步研究裂隙岩含水层污染多分集阵列GPR探测与成像技术.通过多分集阵列发射波形联合优化设计和数据张量分析模型构建,研究多分集阵列GPR认知探测裂隙岩含水层污染方法,系统提出GPR裂隙岩含水层污染探测技术,在GPR探测定量或定性分析方面取得原创性成果.本项目研究成果将为促进我国地下水污染物的监测先进设备研制提供理论基础。

主要针对宽带探地雷达（GPR）地下水污染探测的非线性和不确定性问题,开展了裂隙岩含水层污染管道化目标多分集阵列GPR探测与成像方法研究。主要研究内容包括：频率分集阵列多目标无模糊定位方法、频率分集阵列发射波形调制方法、频率分集阵列近场目标定位张量分析方法、极化频率分集阵列雷达地下目标成像方法、GPR目标图像特征双曲线检测的深度神经网络方法、GPR目标几何尺寸估计的深度神经网络方以及GPR目标介电常数估计的深度神经网络方法等。.研究了随机频率分集阵列的频率增量系数向量和目标定位模糊之间的关系，提出了基于频率增量系数代数关系的判断准则和基于目标位置关系矩阵秩的判断准则，进而对频率增量系数进行约束来实现随机置换频率分集阵列多目标无模糊定位。将FDA与阵元波形调制、多输入多输出（MIMO）、极化等结合进行目标无模糊定位和成像，提出了频偏索引调制FDA发射结构、构建了多分集FDA-MIMO阵列近场探测信号张量模型、建立了频域极化散射矩阵元素与埋地目标走向间的定量关系，并推导了相应的算法。研究了结合GPR物理机制和深度学习网络的GPR目标特征检测和参数估计方法，设计了级联结构的卷积神经网络(CNN)模型，提出了基于方向引导的特征数据补全方法，并通过分类器识别目标特征双曲线。在提取目标特征双曲线基础上，提出了直接根据目标的特征双曲线来估计其几何尺寸大小和基于模型求解和范例学习相结合的目标介电常数估计方法。通过仿真或/和实测数据对相应的理论和模型的有效性进行了验证。在GPR目标探测定量或定性分析方面取得了创新性成果，克服了宽带GPR表层逆散射成像的非线性和不确定性问题,为未来GPR高精度、智能化发展需求提供了理论和技术借鉴。