储层特征参数估算与模糊评价方法研究

发布时间：2018-04-20 17:38

本文选题：储层特征参数估算 + 储层评价　；参考：《中国石油大学(华东)》2015年硕士论文

【摘要】：现代地震数据采集与处理技术的发展为储层目标研究提供了良好的叠前与叠后地震资料,如何有效提取地震资料中的地层岩性和孔隙流体信息成为油藏地球物理研究的难点与热点。以测井揭示的地层岩性与孔隙流体信息为导向数据,研究地震资料中储层目标特征信息的提取与评价方法具有重要的研究意义与工业价值。本研究的目的在于研究出一套叠前和叠后地震属性提取与优选、储层特征参数估算与评价方法及技术,从大量地震资料中有效地提取出刻画储层岩性、物性及含油气性的储层特征参数,评价其在储层目标研究中的适应性与可信度,实现储层目标的定量分析。论文在系统研究与分析了叠前与叠后地震属性理论与提取方法的基础上,引入层次聚类分析-偏最小二乘回归方法,优选出与地层岩性、物性与含流体相关性好的叠前与叠后地震属性。在研究与分析了测井综合解释理论与实际测井数据的基础之上,引入模糊聚类算法挑选出能够用于岩性解释,同时也能够表征储层的自然伽马曲线、密度曲线、声波时差曲线与有效孔隙度曲线作为目标估算的储层特征参数。在研究相关向量机理论和方法的基础之上,通过模型数据对比,阐述了相关向量回归在回归预测性能方面的优势,并对优选出来的地震属性与测井曲线进行预处理,进而构建地震属性与目标储层特征参数之间的非线性相关向量回归模型,估算出各储层特征参数三维数据体。在研究了模糊推理系统理论和方法以及综合分析了储层特征参数与岩心地质资料的基础之上,依据岩性解释的基本原理,构建了语言式规则推理的Mamdani型模糊推理系统,同时详细阐述了模糊推理过程中的参数确定与模糊规则库的修剪,并应用最速下降算法计算规则权重获取最优模糊规则库,综合储层特征参数与岩心资料进行地层岩性解释。针对储层分类的目标,构建了基于T-S模型的模糊推理系统,融合估算的储层特征参数,根据融合结果得分,运用储层分类输出模型获取储层的分类结果。通过模型数据测试与胜利油田埕北工区实际资料应用,层次聚类-偏最小二乘回归地震属性优选方法能够有效地优选出与目标参数最为紧密的、个数较少的地震属性;相关向量回归方法能够很好地拟合出地下地层构造形态、沉积体系,同时与井曲线吻合度较高;基于Mamdani型的模糊推理系统能够较好地解释地层岩性;基于T-S模型的模糊推理系统能够较好地反映出地层中砂体的展布特征,吻合实际钻井结果。
[Abstract]:The development of modern seismic data acquisition and processing technology provides good pre-stack and post-stack seismic data for reservoir target research. How to effectively extract the formation lithology and pore fluid information from seismic data has become a difficult and hot point in reservoir geophysics. Based on the information of formation lithology and pore fluid revealed by logging, it is of great significance and industrial value to study the extraction and evaluation method of reservoir target characteristic information from seismic data. The purpose of this study is to develop a set of prestack and post-stack seismic attributes extraction and optimization, reservoir characteristic parameters estimation and evaluation methods and techniques, and effectively extract from a large number of seismic data to describe reservoir lithology. The reservoir characteristic parameters of physical property and oil-bearing property are evaluated to evaluate their adaptability and credibility in the study of reservoir target, and the quantitative analysis of reservoir target is realized. Based on the systematic study and analysis of prestack and poststack seismic attribute theory and extraction method, the hierarchical cluster analysis and partial least square regression method are introduced to optimize the selection and lithology of strata. Prestack and poststack seismic attributes with good correlation between physical properties and fluids. Based on the research and analysis of the comprehensive interpretation theory and the actual logging data, the fuzzy clustering algorithm is introduced to select the natural gamma curves and density curves that can be used in lithology interpretation. Acoustic moveout curve and effective porosity curve are used as reservoir characteristic parameters for target estimation. On the basis of studying the theory and method of correlation vector machine, the advantage of correlation vector regression in regression prediction performance is expounded through the comparison of model data, and the preprocessing of the selected seismic attributes and logging curves is carried out. Then the nonlinear correlation vector regression model between seismic attributes and target reservoir characteristic parameters is constructed and the 3D data volume of each reservoir characteristic parameter is estimated. On the basis of studying the theory and method of fuzzy inference system and synthetically analyzing reservoir characteristic parameters and core geological data, according to the basic principle of lithology interpretation, a Mamdani fuzzy inference system based on linguistic rule reasoning is constructed. At the same time, the parameter determination and the pruning of fuzzy rule base in the process of fuzzy reasoning are expounded in detail, and the optimal fuzzy rule base is obtained by calculating the weight of rules by using the steepest descent algorithm, and the formation lithology interpretation is carried out by synthesizing the reservoir characteristic parameters and core data. Aiming at the target of reservoir classification, a fuzzy inference system based on T-S model is constructed, and the estimated reservoir characteristic parameters are fused. According to the score of the fusion result, the reservoir classification results are obtained by using the reservoir classification output model. Based on the model data test and the application of actual data in Chengbei area of Shengli Oilfield, the hierarchical clustering and partial least square regression seismic attribute optimization method can effectively select the seismic attributes which are most closely related to the target parameters and the number of seismic attributes is relatively small. The correlation vector regression method can fit the structure and sedimentary system of underground strata very well, at the same time, the fuzzy inference system based on Mamdani model can explain the lithology of strata well, at the same time, the correlation vector regression method can fit well with the well curve. The fuzzy inference system based on T-S model can well reflect the distribution characteristics of sand bodies in the formation and coincide with the actual drilling results.
【学位授予单位】：中国石油大学(华东)
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P618.13

【相似文献】