肖晓,闫建平,郭伟,钟光海,丁明海,罗光东

• 1:西南石油大学计算机科学学院
• 2:西南石油大学地球科学与技术学院
• 3:天然气地质四川省重点实验室(西南石油大学)
• 4:中国石油勘探开发研究院

摘要(Abstract)：

页岩气储层看似均一，实则在矿物、有机质、物性及含气性等方面具有强的非均质性特征，多元回归法和经验公式法等常规方法难以精确计算页岩气储层的甜点参数。机器学习中LightGBM回归算法具有训练效果好、不易过拟合及训练速度快等优点。通过选择自然伽马、无铀伽马、钍钾比、钍铀比、铀、钍、钾、声波时差、中子、密度10条能够反映储层甜点特性的测井曲线数据作为模型输入，储层甜点参数作为输出，开展多测井曲线变量贡献率分析，运用递归特征消除算法获取最优特征子集，最终构建了基于LightGBM回归算法的页岩气储层甜点参数（孔隙度、总有机碳含量、吸附气量、游离气量及总含气量）预测模型。进而对川南地区未参与模型训练的一口页岩气井进行甜点参数预测，计算结果与岩心测试甜点参数数据相关性达到了0.90以上，具有较强的泛化能力，表明LightGBM回归机器学习是一种复杂页岩气储层甜点参数有效、低成本的预测方法。

关键词(KeyWords)： 页岩气储层;甜点参数;总有机碳;吸附气量;游离气量;LightGBM回归算法

Abstract:

Keywords:

基金项目(Foundation): 中国石油-西南石油大学创新联合体科技合作项目（2020CX020000）;; 四川省自然科学基金项目（2022NSFSC0287）;; 高等学校学科创新引智计划（111计划）（D18016）;; 中石油科技部“十四五”重大专项（2021DJ1901）;; 南充市校科技战略合作项目（SXHZ017）联合资助

作者(Author): 肖晓,闫建平,郭伟,钟光海,丁明海,罗光东

参考文献(References)：

• [1]邹才能，董大忠，杨桦，等.中国页岩气形成条件及勘探实践[J].天然气工业，2011,31(12):26-39,125.ZOU C, DONG D, YANG H, et al. Conditions of shale gas accumulation and exploration practices in China[J]. Natural Gas Industry,2011,31(12):26-39,125.
• [2]邹才能，董大忠，王社教，等.中国页岩气形成机理、地质特征及资源潜力[J].石油勘探与开发，2010,37(6):641-653.ZOU C,DONG D,Wang S,et al. Geological characteristics,formation mechanism and resource potential of shale gas in China[J]. Petroleum Exploration and Development,2010,37(6):641-653.
• [3]曾凡武.彭水地区五峰组-龙马溪组页岩气富集主控因素[J].中国煤炭地质，2016,28(11):25-29.ZENG F. Main controlling factors of shale gas enrichment in pengshui area Wufeng,Longmaxi formations[J]. Coal Geology of China,2016,28(11):25-29.
• [4]李玉喜，乔德武，姜文利，等.页岩气含气量和页岩气地质评价综述[J].地质通报，2011,30(S1):308-317.LI Y,QIAO D,JIANG W,et al. Gas content of gas-bearing shale and its geological evaluation summary[J]. Geological Bulletin of China,2011,30(S1):308-317.
• [5]侯华星，欧阳永林，曾庆才，等.四川长宁页岩总有机碳地震定量预测方法[J].东北石油大学学报，2016,40(5):18-27,5.HOU H,OUYANG Y,ZENG Q,et al. Seismic quantitative prediction technique for shale TOC in the Changning,Sichuan[J]. Journal of Northeast Petroleum University,2016,40(5):18-27,5.
• [6]杨钦，苏思远，李昂，等.孔隙类型对页岩气赋存状态的影响——以川南长宁地区五峰组—龙马溪组页岩为例[J].中国矿业大学学报，2022,51(4):704-717.YANG Q,SU S,LI A,et al. Influence of pore type on the occurrence state of shale gas:Taking Wufeng-Longmaxi formation shale in Changning area of southern Sichuan as an example[J]. Journal of China University of Mining&Technology,2022,51(4):704-717.
• [7]孙宁.凤冈一区五峰组—龙马溪组富有机质页岩孔隙特征及含气性研究——以XX井为例[J].中国煤炭地质，2021,33(7):27-33,77.SUN N. Study on Wufeng Formation-Longmaxi formation organic matter rich shale pore features in Fenggang workspace No.1——A Case Study of Well XX.[J]. Coal Geology of China,2021,33(7):27-33,77.
• [8]温康，闫建平，钟光海，等.川南长宁地区五峰组—龙马溪组页岩气评价新方法[J].岩性油气藏，2022,34(1):95-105.WEN K,YAN J,ZHONG G,et al. New method of shale gas evaluation of Wufeng-Longmaxi formation in Changning area,southern Sichuan basin[J].Lithologic Reservoirs,2022,34(1):95-105.
• [9]黄莉莎，闫建平，郭伟等.基于随机森林回归算法的低电阻率页岩气储层饱和度评价[J].测井技术，2023,47(1):22-28.HUANG L,YAN J,GUO W,et al. Evaluation of low resistivity shale gas reservoir saturation based on Random Forest regression method[J].Well Logging Technology,2023,47(1):22-28.
• [10]RICKMAN R,MULLEN M J,PETRE J E,et al. A practical use of shale petrophysics for stimulation design optimization:All shale plays are not clones of the barnett shale[C].//SPE annual technical conference and exhibition. Society of Petroleum Engineers,2008.
• [11]钟光海，谢冰，周肖，等.四川盆地页岩气储层含气量的测井评价方法[J].天然气工业，2016,36(8):43-51.ZHONG G,XIE B,ZHOU X,et al. A logging evaluation method for gas content of shale gas reservoirs in the Sichuan basin[J]. Natural Gas Industry,2016,36(8):43-51.
• [12]张冲，郭聪，朱林奇，等.考虑有机质校正的页岩气储层总孔隙度测井评价方法[J].煤炭学报，2017,42(6):1527-1534.ZHANG C,GUO C,ZHU L,et al. A logging interpretation method for total porosity considering organic matter correction of shale gas reservoirs[J]. Journal of China Coal Society,2017,42(6):1527-1534.
• [13]钟光海，陈丽清，廖茂杰，等.页岩气储层品质测井综合评价[J].天然气工业，2020,40(2):54-60.ZHONG G,CHEN L,LIAO M,et al. A comprehensive logging evaluation method of shale gas reservoir quality[J]. Natural Gas Industry,2020,40(2):54-60.
• [14]钟文俊，熊亮，黎鸿，等.测井评价技术在威荣深层页岩气田中的应用[J].油气藏评价与开发，2021,11(1):38-46.ZHONG W,XIONG L,LI H,et al. Application of log evaluation technology in Weirong deep shale gas field[J]. Petroleum Reservoir Evaluation and Development,2021,11(1):38-46.
• [15]孟召平，郭彦省，刘尉.页岩气储层有机碳含量与测井参数的关系及预测模型[J].煤炭学报，2015,40(2):247-253.MENG Z,GUO Y,LIU W. Relationship between organic carbon content of shale gas reservoir and logging parameters and its prediction model[J]. Journal of China Coal Society,2015,40(2):247-253.
• [16]周广照，李显明，黄斌，等.优化BP神经网络在川西上三叠统陆相页岩含气性预测中的应用[J].矿物岩石，2017,37(3):90-96.ZHOU G,LI X,HUANG B,et al. Application of optimized BP neural networks to gas content prediction of continental shale in upper triassic of westerm sichuan basin[J]. Mineralogy and Petrology,2017,37(3):90-96.
• [17]叶绍泽，曹俊兴，吴施楷，等.基于深度置信网络的总有机碳含量预测方法[J].地球物理学进展，2018,33(6):2490-2497.YE S,CAO J,WU S,et al. Prediction method of total organic carbon content based on deep belief nets[J]. Progress in Geophysics,2018,33(6):2490-2497.
• [18]袁颖，谭丁，于少将，等.基于贝叶斯正则化改进BP神经网络的页岩气有机碳含量预测模型[J].地质与勘探，2019,55(4):1082-1091.YUAN Y,TAN D,YU S,et al. A prediction model for shale gas organic carbon content based on improved BP neural network using bayesian regularization[J]. Geology and Exploration,2019,55(4):1082-1091.
• [19]张少龙，闫建平，郭伟，等.基于岩石物理相的深层页岩气地质—工程甜点参数测井评价方法——以四川盆地LZ区块五峰组—龙马溪组为例[J].石油地球物理勘探，2023,58(1):214-227.ZHANG S,YAN J,GUO W,et al. Logging evaluation method of geological-engineering sweet spot parameters for deep shale gas based on petrophysical facies:A case study of the Wufeng-Longmaxi Formation in LZ block of Sichuan basin[J]. Oil Geophysical Prospecting,2023,58(1):214-227.
• [20]ZHAN Z,YOU Z,LI L,et al. Accurate prediction of ncRNAprotein interactions from the integration of sequence and evolutionary information[J]. Frontiers in Genetics,2018,9(1):458.
• [21]MA X,SHA J,WANG D,et al. Study on a prediction of P2P network loan default based on the machine learning LightGBM and XGboost algorithms according to different high dimensional data cleaning[J]. Electronic Commerce Research and Applications,2018,31(1):24-39.
• [22]冯易，王杜娟，胡知能，等.基于改进LightGBM集成模型的胃癌存活性预测方法[J].中国管理科学，2023,31(10):234-244.FENG Y,WANG D,HU Z,et al. Prediction method for gastric cancer survivability based on an improved LightGBM ensemble model[J].Chinese Journal of Management Science,2023,31(10):234-244.
• [23]KE G,MENG Q,FINLEY T,et al. LightGBM:A highly efficient gradient boosting decision tree[C].//Proc of Conference on Neural lnformation Processing Systems,2017:569-577.
• [24]FRIEDMAN J H. Stochastic gradient boosting.[J]Comput Stat Data An,2002,38:367-378.
• [25]LUO Y,XU Q,LI W,et al. A multi-step decision prediction model based on LightGBM[C].//Proc of International Conference on Big Data,2021:5714-5718.