基于马尔科夫概率模型的碳酸盐岩储集层测井岩性解释

发布时间：2018-02-17 06:11

本文关键词： 苏里格气田奥陶系碳酸盐岩岩性识别马尔科夫概率模型朴素贝叶斯方法转移概率　出处：《新疆石油地质》2017年01期 　论文类型：期刊论文

【摘要】：苏里格气田苏东41-33区块下奥陶统马家沟组马五段碳酸盐岩储集层受沉积、成岩等因素的影响,岩性复杂多样,因此,岩性的识别是储集层评价的关键。测井和录井资料分析表明,苏东41-33区块马五段主要发育灰岩、白云岩、白云质灰岩、灰质白云岩、泥质白云岩、泥质灰岩和泥岩7种岩性。通过敏感性参数分析,选取反映岩性的自然伽马、密度、光电吸收截面指数和补偿中子4种测井参数,采用马尔科夫概率模型约束的朴素贝叶斯方法进行多参数综合解释。通过与测井和录井岩性资料分析对比,识别正确率达到85.34%,相比传统的朴素贝叶斯方法,岩性识别正确率提高12.39%.此方法得到识别精度较高的碳酸盐岩岩性测井解释模型,是一种有效的复杂岩性识别方法。
[Abstract]:The carbonate reservoir of Majiagou formation of Lower Ordovician Majiagou formation in Suodong 41-33 block of Sulige gas field is affected by sedimentation, diagenesis and other factors. The analysis of logging and logging data shows that limestone, dolomite, dolomite, dolomite, mudstone and muddy dolomite are mainly developed in the fifth member of Ma Wu formation in Block 41-33 of East Jiangsu Province. Through sensitivity parameter analysis, four logging parameters, natural gamma, density, photoelectric absorption cross section index and compensated neutron, are selected to reflect lithology. The naive Bayes method with Markov probability model constraint is used to interpret the multi-parameter comprehensive interpretation. By analyzing and comparing with logging and logging lithology data, the recognition accuracy is 85.34, compared with the traditional naive Bayes method. The accuracy rate of lithology recognition is increased by 12.390.This method is an effective method for identifying complex lithology, and the interpretation model of carbonate lithology logging with high recognition accuracy is obtained.
【作者单位】：中国地质大学地球物理与信息技术学院;中国地质大学能源学院;中国石油长庆油田分公司第四采气厂;中国地质大学数理学院;
【基金】：国家科技重大专项(2016ZX05014-001) 国家自然科学基金(41172130,U1403191) 中央高校基本科研项目(2-9-2015-209)
【分类号】：P618.13;P631.81

【参考文献】