【作者】 万玲；

【导师】 林君；

【作者基本信息】 吉林大学 ， 测试计量技术及仪器， 2013， 博士

【摘要】 当地层导电性较强时，电阻率信息对磁共振探测的影响将非常大。因此，在磁共振反演解释中必须考虑地下电阻率分布信息。然而，传统磁共振地层含水量反演多采用均匀半空间模型，忽略电阻率分布信息对结果的影响。针对这一问题，本文基于多层电介质中磁共振响应理论，提出MRS与瞬变电磁（TEM）联合反演方法，通过电阻率分布信息对含水量反演过程的实时修正，提高解释结果的准确度。针对标准遗传算法易未成熟收敛而难以得到全局最优解问题，提出自适应遗传（AGA）反演算法，基于繁殖规则，动态调整交叉概率和变异概率，模型数据验证了AGA反演算法的有效性及联合反演对比MRS单独反演的优势。在此基础上，通过仿真分析电阻率分布信息对MRS正反演的影响，同时为便于野外工程应用，对现有磁共振与瞬变电磁联合探测系统性能和结构进行了改进。最后给出了三个野外实验结果和应用实例。论文的主要研究内容和取得的研究成果如下：1.研究了地表回线源在层状导电介质中产生的磁场分布规律，推导了层状介质模型磁共振响应表达式。通过不同地层电阻率模型磁共振响应分析，得出100米以下的深层电阻率信息依然会影响磁共振信号的结论。通过对不同浅层电阻率模型的磁共振响应对比分析，得出必须改进瞬变电磁仪器性能，减少探测盲区的结论。2.以标准遗传算法为基础，提出了一种自适应遗传反演算法。利用模型参数的最小分辨率及搜索空间确定染色体长度；利用群体中个体离散程度和平均适应度自适应调整个体交叉、变异概率；通过综合考虑个体目标函数值及所在群体的分散程度设计了适应度函数。将AGA反演算法应用于磁共振、瞬变电磁单一数据反演，验证反演算法的有效性。3.提出了磁共振与瞬变电磁联合反演方法。详细介绍了正则化反演方法，并给出了一种兼顾反演结果稳定性与算法计算效率的正则化参数自动调整方法。建立了具有光滑模型约束的MRS与TEM联合反演目标函数，依据Archie定律，进一步缩小模型参数搜索空间，降低反演问题的多解性。最后详细设计了基于AGA反演算法的MRS与TEM联合反演方案。联合反演优势在均匀半空间模型、层状模型反演算例中均得以体现。在理论信号中分别加入不同信噪比噪声，反演结果稳定。且MRS-TEM联合反演方法即使在加入10%噪声情况下，结果仍能较真实地反映地下含水单元模型结构，相对于传统MRS单独反演，优势明显。4.根据现有地面磁共振与瞬变电磁联合探测系统组成结构，针对磁共振采集信号死区时间长和瞬变电磁发射关断时间慢等问题，在磁共振系统发射单元采用“Q-Switch”技术，缩短了探测死区时间，提高信噪比。研究了瞬变电磁半谐振吸收电路，实现了发射电流的快速关断，降低了瞬变电磁浅部探测盲区。5.本文研究的磁共振与瞬变电磁联合反演方法在内蒙古正镶白旗水源地、吉林长春烧锅实验场和辽宁绥中沿海地区海水入侵探测工程中得到应用，实现了对地下水的定量定位勘查。联合反演得到的地层结构和含水单元结构可实现相互验证，且探测结果均得到了钻探资料的支持，充分验证了本文研究方法的准确性和实际意义。本文创新性工作主要体现为如下几个方面：1.提出了磁共振与瞬变电磁联合反演方法，构建了加权联合问题目标函数，解决了以均匀半空间为模型所导致解释精度低的问题。2.提出了一种适用于联合反演的自适应遗传算法，解决了标准遗传算法易陷入局部极值的问题，显著地提高了反演结果的准确度。3.对联用仪中的Q-switch、半谐振吸收方法进行了改进设计，解决了死区和关断时间长对反演解释的影响。更多还原

【Abstract】 When the underground is conductive, the electrical conductivity will have a great effecton the Magnetic Resonance Sounding (MRS) signal and can not be ignored in the inversion.The traditional interpretation schemes of MRS used to be limited to the homogeneoushalf-space and take no account of the resistivity in the inversion, in turn resulting in theambiguous aquifer boundaries. Based on the MRS theory in the multi-layer earth model, wedeveloped a new scheme for joint inversion of MRS data and TransientElectromagnetic(TEM) data. By correcting MRS inversion with resistivity in real time, theaccuracy of inversion result was improved greatly. The Adaptive Genetic Algorithm(AGA)was applied. By adaptively adjusting the probability of crossover and mutation, AGA has notonly avoided the premature convergence of GA but also has obtained the global optimalsolution. On this basis, by simulating the effect of conductivity in forward and inversion ofMRS and for convenience in field experiment,we improved the performance and structureof MRS-TEM associated detection system. Finally three field experiments and applicationexamples are given. The main research contents and results are as follows:1. We study the magnetic field distribution produced by the surface gyrus line in layeredconductive medium and derive the magnetic resonance response expression of a layeredmedium model. By analyzing the responses of different layered resistivity models, aconclusion can be obtained that the resistivity information below100metres will stillinfluence the magnetic resonance signal. Among numerous geophysical techniques,MRS-TEM associated detection is the best solution. Comparing the magnetic resonanceresponses of different shallow resistivity models, it is concluded that the measureimproving TEM instrument performance must be taken to reduce detection blind area.2. In view of the standard genetic algorithm (GA),the adaptive genetic inversion algorithmwas put forward. According to the minimum resolution and the search space of modelparameters to determine the length of the chromosome; According to the individualdispersion and the average fitness to adaptively adjust the crossover and mutationprobability. By considering the value of the individual objective fuction and thedispersion of the group, the fitness function was designed. Using the AGA in the inversion of MRS and TEM data respectively, to vertify the effeciency of the algorithm.3. A method of joint inversion based on MRS–TEM data was put forward, which introduce the regularization inversion in detail and presents a parameter automatic adjustment method combining the stability of inversion results with computational efficiency. An objective function of MRS and TEM joint inversion is established with smooth model constraints and further narrows the searching space and reduces the uncertainty of inverse problem according to Archie’s law. At last, the MRS and TEM joint inversion scheme are designed in detail based on the AGA algorithm.The advantages of joint inversion is reflected in uniform half space models and layered models. Moreover, mixing noise of different SNR in the theoretical signal, the inversion results is still stable and reflect underground aquifer structure, even in the case of adding10%noise. Compared with the traditional MRS single inversion, the advantage is obvious.4. Considering the existing MRS-TEM associated system structure and aiming at the longdead-time of MRS and long turn-off time of TEM, a ’Q-Switch’ technique was used inMRS transmitting system to shorten the dead-time and improve the signal-to-noise ratio,a half-resonant&absorbing circuit was realized to shorten the TEM turn-off time anddecrease the TEM blind area.5. Joint inversion of MRS and TEM in this paper has been applied in the field to realizethe quantitative and qualitative imaging of groundwater, like BaiQi, Changchun, andLiaoning. The formation structure and the water unit structure obtained by the jointinversion can reach mutual authentication and are also supported by drilling data, bywhich the accuracy and the practical significance of the inversion method has been fullyverified.The main innovation lies as follow:By analyzing the impact of different resistivity distribution on MRS response, wefound that the resistivity information in the depths will still influence the MRSsignal. So I proposed that MRS and TEM associated detection method is the bestsolution among numerous geophysical techniques. The adaptive genetic inversion algorithm(AGA) I proposed can solve the slowconvergence speed and easy to fall into local optimal solution problem of thestandard genetic algorithm (GA). Use the AGA as the inversion algrithom ofJoint inversion of MRS and TEM for the first time, both the theoretical and fielddata vertified the effeciency of the algorithm.Joint inversion of MRS and TEM data was proposed. According to characteristicsof MRS and TEM response, the normalized objective function of joint inversionwas established, and the mutual constraints between the model parameters wererealized based on archie formula. By correcting the inversion process in real timeusing the resistivity distribution information, the accuracy of MRS interpretationresults can be improved greatly.The MRS and TEM detection system was improved by making the effect ofresistivity distribution on MRS response clear. A half-resonant&absorbingcircuit was realized to shorten the TEM turn-off time and decrease the TEM blindarea. A ’Q-Switch’ technique was used in MRS transmitting system to shortenthe dead-time and improve the signal-to-noise ratio.更多还原