基于马尔科夫-贝叶斯模拟算法的多地震属性沉积相建模方法——以苏里格气田苏10区块为例

发布时间：2018-06-17 00:27

本文选题：沉积相建模 + 地震属性　；参考：《油气地质与采收率》2017年03期

【摘要】：在钻井资料较少的气田早期评价阶段,仅以钻井数据建立的沉积相模型存在很大的不确定性,难以满足地质研究的精度要求。为此,在沉积相建模过程中引入多地震属性作为约束条件。首先在贝叶斯模型框架约束下建立先验概率分布,然后根据模拟点周围的钻井数据和地震数据将先验概率分布更新为后验概率分布,在更新过程中运用马尔科夫假设解决不同变量间交叉矩阵的不稳定问题,最终通过对模拟点的后验概率分布进行随机抽样,从而获取其沉积相类型。以苏里格气田北部苏10区块为研究对象,通过分析地震属性与沉积相的关系,优选出均方根振幅、平均瞬时频率、有效频带和衰减因子4种地震属性,针对在整合多变量时互协方差计算量大的问题,运用马尔科夫-贝叶斯模拟算法,对多地震属性进行融合,得到多变量融合概率场信息,进而建立沉积相模型。研究结果表明,模拟结果与人工编绘的沉积相分布规律的吻合率超过80%,通过交叉检验分析预测误差在1%以内。
[Abstract]:In the early evaluation stage of gas fields with less drilling data, there is great uncertainty in the sedimentary facies model only established by drilling data, and it is difficult to meet the precision requirements of geological research. Therefore, multiple seismic attributes are introduced as constraints in sedimentary facies modeling. A priori probability distribution is established under the constraints of the Bayesian model framework, and then the prior probability distribution is updated to a posteriori probability distribution according to the drilling data and seismic data around the simulated point. In the process of renewal, Markov hypothesis is used to solve the instability problem of cross matrix between different variables. Finally, the types of sedimentary facies are obtained by random sampling of posteriori probability distribution of simulated points. Taking Su10 block in the northern part of Sulige gas field as the research object, through analyzing the relationship between seismic attributes and sedimentary facies, four kinds of seismic attributes, such as mean square amplitude, mean instantaneous frequency, effective frequency band and attenuation factor, are selected. In order to solve the problem of large amount of calculation of cross covariance in the integration of multi-variables, Markov Bayesian simulation algorithm is used to fuse the multi-seismic attributes, and the multi-variable fusion probability field information is obtained, and then the sedimentary facies model is established. The results show that the distribution rate of the simulated results is more than 80%, and the prediction error is less than 1%.
【作者单位】：中国地质大学(北京)地球物理与信息技术学院;中国石油长庆油田分公司第三采气厂;中国地质大学(北京)数理学院;中国地质大学(北京)能源学院;
【基金】：国家科技重大专项“大型油气田及煤层气开发”(2016ZX05014-001)
【分类号】：P618.13;P631.4

【相似文献】