长江大学学报（自然科学版）2025，Vol.22Issue(2)：18-27,10.
测录井融合潜山储层复杂岩性随机森林识别方法
Comprehensive lithology identification method for complex reservoir of buried hill based on wireline and cutting logs:A case study of Huizhou 26-6 structure in the Pearl River Mouth Basin
刘君毅 1王清辉 1冯进 1管耀 1刘杰 1宋伟 1赵鹏 1王显南1
作者信息
- 1. 中海石油(中国)有限公司深圳分公司,广东 深圳 518054||中海石油深海开发有限公司,广东 深圳 518054
- 折叠
摘要
Abstract
Accurate lithology identification of buried hill strata is the basis of fine reservoir characterization.Many different kinds of lithologies were encountered while drilling in the buried hill section of Huizhou 26-6 structure in the eastern Pearl River Estuary Basin,such as tectonic schist,mylonite,cataclastic granite,diorite,granodiorite,granite and altered ophiase.The accuracy of lithology identification solely based on conventional logging or elemental logging information is low and cannot meet the production needs.The article proposes to organically integrate the two types of information.Firstly,based on the analysis results of core experiments,the elemental logging data is corrected to eliminate systematic errors,and the corrected data is verified to be applicable for quantitative lithology identification through comparison with elemental logging examples.Then,based on the identification results of core thin sections from samples of different rock types,the logging response characteristics of different rock types were analyzed,and the sensitivity of different curves was analyzed.Five logging curves including natural gamma,resistivity,neutron porosity,density,and acoustic time difference were determined,as well as six element logging curves including silicon,aluminum,calcium,iron,magnesium and potassium as input curves for the next step of rock identification method.Finally,based on the characteristics of the data in the study area,the random forest machine learning algorithm was selected and trained and tested on samples.The results showed that the accuracy of lithology recognition using comprehensive logging and logging information reached 92.6％,which is more than 20 percentage points than using logging or logging information alone before.The actual data processing of new drilling shows that this method can effectively solve the problem of identifying complex lithology in the target layer of the study area.The comparative analysis of multiple well connected profiles further verifies that the processing results comply with regional geological laws.关键词
惠州凹陷/潜山/岩性识别/常规测井/元素录井/随机森林Key words
Huizhou sag/buried hill/lithology identification/conventional logging/element logging/random forest引用本文复制引用
刘君毅,王清辉,冯进,管耀,刘杰,宋伟,赵鹏,王显南..测录井融合潜山储层复杂岩性随机森林识别方法[J].长江大学学报（自然科学版）,2025,22(2):18-27,10.
