测井技术2023，Vol.47Issue(6)：671-678,8.DOI:10.16489/j.issn.1004-1338.2023.06.004
基于随机森林算法的深层低对比度气藏流体识别
Fluid Identification of Deep Low-Contrast Gas Reservoirs Based on Random Forest Algorithm
摘要
Abstract
In order to solve the problems such as unclear formation mechanism and poor fluid identification effect in deep gas reservoirs with low contrast of Bashijiqike formation in Bozi well area,Tarim basin,the mechanism of formation with low contrastis deeply analyzed based on the analysis data of cast thin section,high pressure mercury injection and nuclear magnetic resonance experiment.Combined with logging and production dynamic data,fluid sensitive factors such as porosity,resistivity,volume modulus,fluid compression coefficient,fluid index,equivalent fluid volume modulus and fluid volume modulus are selected to identify fluids by random forest algorithm.The results show that the low contrast formation mechanism is different in the region.The reservoirs with low contrast in the southern well area is the result of the combination of formation water salinity,reservoir physical property and pore structure.However,the degree of carbonate cement development is the main factor of the reservoirs with low contrast in the northern well area.The accuracy of the fluid identification model of low contrast gas reservoir based on random forest algorithm is 89.25%,which weakens the multiple solutions caused by a single fluid identification factor and provides a reliable basis for the efficient development of gas fields.关键词
流体识别/深层/随机森林/低对比度/博孜井区Key words
fluid identification/deep reservoir/random forest/low contrast/Bozi well area引用本文复制引用
曹原,赵元良,袁雪花,袁龙,荣俊卿,赵盼,别康..基于随机森林算法的深层低对比度气藏流体识别[J].测井技术,2023,47(6):671-678,8.基金项目
国家十三五重大科技专项"陆相页岩气层测井评价技术"(2017ZX05039-002) （2017ZX05039-002）
中国石油集团测井有限公司十大科技项目"重点勘探领域测井解释评价核心技术攻关"(CNLC2022-08B02) （CNLC2022-08B02）
