松辽盆地砂岩型铀矿钻孔岩性的测井识别  

Logging identification of borehole lithology of sandstonetype uranium deposit in Songliao Basin

作　　者：段忠义 肖昆 杨亚新[1] 黄笑 王殿学 徐艺宸 焦常伟 DUAN ZhongYi;XIAO Kun;YANG YaXin;HUANG Xiao;WANG DianXue;XU YiChen;JIAO ChangWei(East China University of Technology,Key Laboratory of Nuclear Resources and Environment,Nanchang 330013,China;Nuclear Industry Group No.243,Chifeng 024000,China)

机构地区：[1]东华理工大学核资源与环境国家重点实验室,南昌330013 [2]核工业二四三大队,赤峰024000

出　　处：《地球物理学进展》2023年第6期2490-2501,共12页Progress in Geophysics

基　　金：江西省主要学科学术和技术带头人培养计划(20204BCJ23027);核资源与环境国家重点实验室联合创新基金(2022NRE-LH-18);江西省自然科学基金(20232BAB203072);东华理工大学研究生创新专项资金项目(DHYC-202315)联合资助。

摘　　要：岩性识别是铀矿勘探和开发中关键环节之一,通过对钻孔中地球物理测井数据的分析和综合处理,可以进一步了解地下岩石的性质、组成和分布情况,为铀矿测井识别提供重要的技术支持.针对北方松辽盆地典型砂岩型铀矿钻孔的综合测井和岩心编录资料,分别采用交会图法和梯度提升树(Gradient Boosting Decision Tree, GBDT)算法对研究区的地层开展了岩性识别研究.结果表明,研究区内有粗砂岩、粉砂岩、细砂岩、中砂岩、黏土、泥岩、砂砾岩7种岩性,利用交会图仅能识别黏土和砂砾岩,不能有效定性划分地层岩性;利用GBDT算法构建的岩性识别模型,对砂岩型铀矿地层岩性识别准确率达到98.52%.所建立的机器学习模型能够有效划分研究区地层岩性,对松辽盆地砂岩型铀矿的测井评价工作具有重要意义.Lithology identification is one of the key links in uranium exploration and development.Through the analysis and comprehensive processing of borehole logging data,the properties,composition and distribution of underground rocks can be further understood,which provides important technical support for uranium ore logging identification.According to the logging core records of typical sandstone uranium holes in the northern Songliao Basin,crossplot method and Gradient Boosting Decision Tree(GBDT)algorithm were respectively adopted to study lithology identification of strata in the study area.The results show that there are seven lithology types in the study area:coarse sandstone,siltstone,fine sandstone,medium sandstone,clay,mudstone and conglomerate,and only clay and conglomerate can be identified by crossplot,which can not be qualitatively classified effectively.According to the lithology identification model constructed by GBDT algorithm,the lithology identification accuracy reaches 98.52%in sandstone type uranium deposit formation.The established machine learning model can effectively classify the formation lithology in the study area,which is of great significance for the logging evaluation of sandstone-type uranium deposits in Songliao Basin.

关 键 词：砂岩型铀矿 测井 岩性识别 交会图 GBDT 

分 类 号：P631[天文地球—地质矿产勘探]