基于随机森林的火山岩岩性测井识别——以准噶尔盆地滴西地区石炭系为例  

Log-based lithology identification of volcanic rocks using random forest method:A case study of Carboniferous strata in the Dixi area,Junggar Basin

作　　者：尚亚洲 张兆辉 许多年[2] 赵雯雯 陈华勇 韩海波 SHANG Ya-Zhou;ZHANG Zhao-Hui;XU Duo-Nian;ZHAO Wen-Wen;CHEN Hua-Yong;HAN Hai-Bo(School of Geological and Mining Engineering,Xinjiang University,Urumqi 830047,China;Research Institute of Petroleum Exploration and Development-Northwest(NWGI),PetroChina,Lanzhou 730020,China;Research Institute of Exploration and Development,PetroChina Xinjiang Oilfield Company,Karamay 834000,China)

机构地区：[1]新疆大学地质与矿业工程学院,新疆乌鲁木齐830047 [2]中国石油勘探开发研究院西北分院,甘肃兰州730020 [3]中国石油股份有限公司新疆油田勘探开发研究院,新疆克拉玛依834000

出　　处：《物探与化探》2024年第4期1025-1036,共12页Geophysical and Geochemical Exploration

基　　金：新疆维吾尔自治区“天池英才”计划项目(51052300560);新疆大学博士启动基金项目(620322016)。

摘　　要：火山岩岩性的准确识别是火山岩油气藏高效勘探开发的重要基础工作。火山岩储层岩性种类多、纵向多期叠置、横向相变快,致使交会图版法对火山岩储层岩性的识别准确率较低。本文在利用网格搜索和正交试验法确定模型最优参数组合基础上,量化评价了常规测井曲线对火山岩岩性的检测能力,将自然伽马、补偿中子、声波时差和地层电阻率作为岩性指示因子,采用随机森林方法建立了准噶尔盆地滴西井区石炭系火山岩岩性的智能识别模型。通过识别研究区5口取心井累计870 m钻井取心段的岩性,识别结果与岩石薄片鉴定结果的符合率达到76.67%,与钻井取心描述结果的符合率高达85.98%,取得了良好的识别效果,为该地区火山岩油气藏的精细评价奠定了基础。The accurate lithologyidentification of volcanic rocksserves as a significant foundation for the efficient exploration and exploitation of volcanic reservoirs.However,volcanic reservoirs exhibit intricate lithologies,longitudinalmultistagesu perimposition,and fast transverse phase transition,which reduce the accuracy of crossplots in lithologyidentification ofvolcanic reservoirs.Based on the optimal parameter combination of the model determined through grid search and orthogonal experiments,this study quantitatively evaluatedthe effects of conventional log curves on the lithologyidentification of volcanic rocks.Withthe natural gamma ray,compensated neutron,sonic interval transit time,and formation resistivity as lithologic indicators,this study builtan intelligent model for the lithology identification of Carboniferous volcanic rocks in the Dixi area in the Junggar Basin using therandom forest method.This study identified the lithologies of thecored intervalswith a cumulative thickness of 870 m infive cored wells in the study area,with the coincidence ratesof the identification results with thin section identification results and core description resultsreaching 76.67%and 85.98%,respectively.This suggestssignificant identification effects.Therefore,this studysets the stagefor the fine-scale evaluation of volcanic reservoirs in the study area.

关 键 词：随机森林 岩性识别 火山岩 机器学习 

分 类 号：P631[天文地球—地质矿产勘探]