基于马蒙算法的反演及其在储集层预测中的应用  

Inversion Based on Markov Chain and Monte-Carlo Method and Application in Reservoir Prediction

作　　者：黄振凯[1] 李占东[2] 梁金中[1] 侯梅芳[1] 

机构地区：[1]中国石油勘探开发研究院,北京100083 [2]东北石油大学石油工程学院,黑龙江大庆163318

出　　处：《新疆石油地质》2012年第6期733-735,共3页Xinjiang Petroleum Geology

基　　金：中石油科技攻关项目(2011A-0203);中石油科技创新基金研究项目(2011D-5006-0304)

摘　　要：论述了基于马尔科夫链-蒙特卡洛算法反演的主要原理,重点阐述了该方法的3个关键环节,即测井资料均一化处理、地震子波提取及合成地震记录和构造建模约束马蒙算法反演。应用结果表明,该方法能够提高砂泥岩薄互层条件下砂体识别的准确程度,更丰富地显示断层两侧砂体发育的细节。同时该方法也有助于准确定位河道砂边界和明确主力砂体之间的接触关系,从而为后期剩余油挖潜和井网结构调整提供依据。The main principle of inversion procedure is presented based on the Markov chain and Monte-Carlo method （MCMC）. Three key links, such as homogenization of logging data, seismic wavelet acquisition, synthetic seismogram and structural modeling and constraint MCMC inversion, are highlighted. The applied results indicate that using this method can improve the accurate degree of identification of sand body in thin sand-mud interbed and enrich the details of sand-body distribution in two flanks on fault. Also, this method is helpful to get an exact location to the boundary of channel sand and determine the contact relationship between the major sand bodies, hence provid- ing basis for tapping potential of the remaining oil and well pattern structural adjustment in late period of an oilfield development.

关 键 词：马尔科夫链-蒙特卡洛算法 地质统计学 贝叶斯判别准则 地震反演 储集层预测 

分 类 号：P631.445.91[天文地球—地质矿产勘探]