【作者】 班丽；

【导师】 刘展；

【作者基本信息】 中国石油大学 ， 地质资源与地质工程， 2009， 博士

【摘要】 高精度重磁勘探要求重磁解释向定量化深入,重磁反演逐步发展到三维定量反演阶段。物性反演以物性变化勾画场源范围,具有模拟复杂地质体的能力和较强的适应能力,是提高和深化重磁解决地质问题能力的重要途径,逐渐成为重磁定量反演的主要方式。本论文基于物性模型进行约束反演方法研究,针对三维定量反演多解性严重和计算量巨大的难点问题,提出了基于重磁场特征和场源形态相关关系的相关约束定量反演方法,对约束机制、反演算法以及约束机制和反演算法的结合方式等反演方法的各个关键问题进行了系统研究。以对高维解空间降维为目标进行约束机制提取方式的研究。对重磁异常局部特征和场源形体参数的相关关系进行了定量研究,提出以视深H s(ΔEΔE_z)反映场源深度,并依据定量相关关系式实现了对场源纵向分布搜索;对于存在先验信息的情况,提出以视深H s反映先验信息控制范围,实现了先验信息约束扩展搜索,建立了一种基于场源分布自动搜索和先验信息约束扩展的相关约束机制。该约束机制的基本思想是对解空间降维,将待定量反演计算的剖分单元数目减小至最小,从而减少反演计算的多解性和计算量。这种约束方式便于和任意反演算法结合,具有很好的适应性。选取区间搜索全局优化算法进行反演计算,对低维区间搜索算法进行高维扩展。针对高维解空间计算量巨大,实际中难以实现的问题,引入概率成像的思想对其进行改进,对剖分单元在各自窗口范围内计算其成像概率,按照成像概率从大到小的顺序对剖分单元进行搜索,使算法达到实用化水平。与其它全局优化算法相比,该算法涉及参数少,选取原则易于掌握。研究了相关约束机制和区间搜索定量反演算法的结合方式,建立了相关约束定量反演方法。首先利用约束机制确定待反演计算的剖分单元,然后分别计算这些剖分单元的成像概率,按照成像概率从大到小的顺序,在各自的物理约束区间内进行区间逼近搜索确定全局极值。设计理论地质模型进行方法试验研究,测试了输入参数的选取原则和反演方法对各种重磁场源及其组合的反演效果。最后应用本研究方法对惠民凹陷临商地区重磁异常进行了视密度和视磁化强度反演,推断了研究区火成岩的空间分布。更多还原

【Abstract】 With the gravity and magnetic (GM) exploration developing, 3D GM inversion has become a trend. Inversion for physical properties which outline source scope according to physical properties chages become an important inversion mode because it can simulate more complex source.However, ambiguous solutions is very serious in 3D GM inversion for physical properties. The huge calculation leads to some quantitative calculation methods which have good inversion results for models can not reach the level of utility. In this thesis, quantitative inversion method for physical properties with correlation constrains which are established based on the relationship between GM anomaly and field source physical parameters is propoesd in order to improve the speed of inversion computation and reduce ambiguous solutions. Severval keys such as constraint intruduction, inversion algorithm and the combination of constrains are studied.Correlation constraint is studied which aim is reducing dimensions of solution space. Virtual depth is proposed to reflect the depth of field sources and establish a quantitative relationship between virtual depth and actual depth of source. Based on the relationship between GM anomalies and field source physical parameters, automatically source searching method is proposed. If there are known information got from well data, information expansion mechanism based on the feature of virtual depth is posed.The nature of these constrains is reducing the number of units which are inversed.These constrains can be combined with any other inversion algorithm easily and have strong adaptability.Interval search algorithm is selected as inversion algorithm. The expansion of inverval algorithm from low dimension to high dimension is achieved.The huge computation of high dimensional searching is beyond now computing ability. The idea of probability tomography is introduced to reduce its computation. Tomography probability of unit is calculated and the order of interval searching is from large to small probability. Compared with other global optimization algorithms, this method involves fewer parameters, the selecting principles are easy to grasp.Quantitative GM inversion under the correlation constraint is studied based on above research results. Firstly, determine which units are inversed according to the correlation constraint mechanism. Then calculate tomography probability of units and search optimal solution in their physical parameter interval from the unit of largest probability firstly.A great deal of theoretical models is designed to summarize the principle to choose parameters and verify the effectiveness of method.Compared with non-constrained inversion results, not only greatly improved computing speed, ambiguous solutions also have been significantly reduced.Finally, the method was applied to inverse virtual density and magnetization of igneous rock in Linshang area of Huimin depression.From the imaging result, the space distribution of igneous rock of the area was deduced.更多还原