水平井井地模型人工智能预测方法研究与应用

doi:10.13809/j.cnki.cn32-1825/te.2025.05.015

摘要/Abstract

摘要：

水平井钻探已成为油公司提高致密与非常规油气单井产量的重要手段，但由于水平井井眼轨迹与地层的空间关系复杂，传统直井分析思路无法有效应用，准确描述水平井井眼轨迹与目的层及围岩的空间组合关系是水平井测井解释的首要任务。基于导眼井构建地层初始模型，利用测井资料正演约束逐段调整模型是目前主流做法，但该方法时效性低，针对同一地区不同井都需要大量重复的正演计算。因此，在水平井测井数据的处理和解释中，建立合理的井地模型是关键。井地模型能准确描述井眼与地层界面之间的空间关系，包括井眼位置与地层界面距离，井轴位置与地层法线方向的夹角等。同时，基于机器学习和人工智能技术的测井数据分析方法，通过智能模型的训练，已经应用于测井数据解释的各个方面，借助人工智能技术有望突破传统方法的瓶颈。因此，提出一种基于多模型集成与深度神经网络的自动化水平井测井解释方法：首先构建包含不同井眼轨迹与地层组合关系的理论模型，生成测井响应样本库；然后整合极端梯度提升树（eXtreme Gradient Boosting，XGBoost）、轻量级梯度提升机（Light Gradient Boosting Machine，LightGBM）、分类提升机（Categorical Boosting，CatBoost）等机器学习模型，通过多层感知器（Multi-Layer Perceptron，MLP）进一步融合各模型的预测结果；最后对实际测井资料开展井轨迹与围岩几何关系的智能自动判识。实例分析显示，该方法在准确捕捉水平井复杂测井响应特征的同时，显著提高了解释速度和精度，能够适应相似地质环境下多口井的快速分析需求，为水平井测井解释提供了一种高效的智能化手段。

关键词: 水平井, 测井解释, 人工智能, 深度学习, 地层建模

Abstract:

Horizontal well drilling has become an important method for oil companies to enhance single-well production in tight and unconventional oil and gas reservoirs. However, due to the complex spatial relationship between the wellbore trajectory of horizontal wells and the formation layers, traditional vertical well analysis methods cannot be effectively applied. Accurately describing the spatial combination relationship between the wellbore trajectory, the target layer, and the surrounding rock is a primary task in horizontal well logging interpretation. To address this issue, the mainstream approach is to construct an initial stratigraphic model based on a pilot well and then adjust the model segment by segment using forward modeling constraints from logging data. However, this process is time-consuming and requires numerous repetitive forward modeling calculations for different wells in the same area. Therefore, in the processing and interpretation of horizontal well logging data, developing a reasonable well-formation model is essential. This model enables an accurate description of the spatial relationship between the wellbore and the formation interfaces, including the distance between the wellbore position and formation interfaces and the angle between the wellbore axis and the formation normal direction. At the same time, logging data analysis methods based on machine learning and artificial intelligence (AI) technologies have been applied to various aspects of logging data interpretation by training intelligent models. With the support of AI technologies, it is expected to overcome the bottlenecks of traditional methods. To this end, the study proposed an automated horizontal well logging interpretation method based on multi-model integration and deep neural networks. First, a theoretical model was constructed incorporating different wellbore trajectories and formation combination relationships, and a logging response sample library was generated. Then, machine learning models such as eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Categorical Boosting (CatBoost) were integrated, and their prediction results were further fused using a multi-layer perceptron (MLP). Finally, intelligent automatic recognition of the geometric relationship between the well trajectory and the surrounding rock was carried out using actual logging data. Case analysis showed that this method accurately captured the complex logging response characteristics of horizontal wells while significantly improving interpretation speed and accuracy. The proposed method meets the demand for rapid analysis of multiple wells in similar geological environments and provides an efficient, intelligent approach to horizontal well logging interpretation.

Key words: horizontal wells, well logging interpretation, artificial intelligence, deep learning, formation modelling

中图分类号:

TE243.1

李裕涛,李潮流,魏兴云, 等. 水平井井地模型人工智能预测方法研究与应用[J]. 油气藏评价与开发, 2025, 15(5): 858-871.

LI Yutao,LI Chaoliu,WEI Xingyun, et al. Research and application of artificial intelligence-based prediction method for horizontal well-formation modelling[J]. Petroleum Reservoir Evaluation and Development, 2025, 15(5): 858-871.

图/表 15

图1

表1

图2

表2

图3

表3

表4

图4

表5

图5

图6

图7

图8

图9

图10

参考文献 42

[1]	潘继平. 中国油气勘探开发新进展与前景展望[J]. 石油科技论坛, 2023, 42(1): 23-31.
	PAN Jiping. New progress and outlook of China’s oil and gas exploration and development[J]. Petroleum Science and Technology Forum, 2023, 42(1): 23-31.
[2]	庞驰彧, 邵昕, 寇宇静. 水平井测井评价方法及应用探究[J]. 石化技术, 2022, 29(8): 88-90.
	PANG Chiyu, SHAO Xin, KOU Yujing. Exploration on logging evaluation method and application of horizontal wells[J]. Petrochemical Industry Technology, 2022, 29(8): 88-90.
[3]	陈猛, 谢韦峰, 张煜, 等. 水平井油水两相流阵列电磁波持水率计算方法及应用[J]. 油气藏评价与开发, 2023, 13(4): 505-512.
	CHEN Meng, XIE Weifeng, ZHANG Yu, et al. Methods and application for water holdup calculation and flowing image based on array electromagnetic wave instrument in horizontal water-oil wells[J]. Petroleum Reservoir Evaluation and Development, 2023, 13(4): 505-512.
[4]	卢比, 胡春锋, 马军. 南川页岩气田压裂水平井井间干扰影响因素及对策研究[J]. 油气藏评价与开发, 2023, 13(3): 330-339.
	LU Bi, HU Chunfeng, MA Jun. Influencing factors and countermeasures of inter-well interference of fracturing horizontal wells in Nanchuan shale gas field[J]. Petroleum Reservoir Evaluation and Development, 2023, 13(3): 330-339.
[5]	邱小雪, 钟光海, 李贤胜, 等. 不同井斜页岩气水平井流动特征的CFD模拟研究[J]. 油气藏评价与开发, 2023, 13(3): 340-347.
	QIU Xiaoxue, ZHONG Guanghai, Xiansheng I, et al. CFD simulation of flow characteristics of shale gas horizontal wells with different inclination[J]. Petroleum Reservoir Evaluation and Development, 2023, 13(3): 340-347.
[6]	刘伟男, 张超谟, 朱林奇, 等. 页岩气水平井TOC测井评价新方法[J]. 物探与化探, 2021, 45(2): 423-431.
	LIU Weinan, ZHANG Chaomo, ZHU Linqi, et al. A new method for TOC logging evaluation in shale gas for horizontal well[J]. Geophysical and Geochemical Exploration, 2021, 45(2): 423-431.
[7]	汪中浩, 易觉非, 赵乾富, 等. 水平井测井资料地质解释应用[J]. 江汉石油学院学报, 2004, 26(3): 70-72.
	WANG Zhonghao, YI Juefei, ZHAO Qianfu, et al. Application of geologic interpretation of horizontal well logging data[J]. Journal of Jianghan Petroleum Institute, 2004, 26(3): 70-72.
[8]	吕萍, 张永敏. 水平井咨询系统的原理与应用[J]. 测井技术, 2004, 28(5): 455-457.
	LU Ping, ZHANG Yongmin. Theory and application of horizontal well advisement system[J]. Well Logging Technology, 2004, 28(5): 455-457.
[9]	胡松, 周灿灿, 王昌学, 等. 泥页岩油气藏水平井评价对策与实践[J]. 地球物理学进展, 2013, 28(4): 1877-1885.
	HU Song, ZHOU Cancan, WANG Changxue, et al. The strategies ad practice of horizontal well evaluation in shale hydrocarbon reservoir[J]. Progress in Geophysics, 2013, 28(4): 1877-1885.
[10]	苏林帅, 蔡明, 郑占树, 等. 井眼扩径对水平井声波测井响应影响的数值模拟[J]. 物探与化探, 2022, 46(2): 467-473.
	SU Linshuai, CAI Ming, ZHENG Zhanshu, et al. Numerical simulation of the effects of borehole enlargement on sonic logging response of horizontal wells[J]. Geophysical and Geochemical Exploration, 2022, 46(2): 467-473.
[11]	LIU J J, LIU J C. Integrating deep learning and logging data analytics for lithofacies classification and 3D modeling of tight sandstone reservoirs[J]. Geoscience Frontiers, 2022, 13(1): 101311.
[12]	LIAO W, GAO C, FANG J, et al. A TCN-BiGRU density logging curve reconstruction method based on multi-head self-attention mechanism[J]. Processes, 2024, 12(8): 1589.
[13]	范翔宇, 孟凡, 赵鹏斐, 等. 一种基于自动机器学习的声波测井曲线重构方法: CN118534544A[P]. 2024-08-23.
[14]	刘军, 钟洁, 倪振, 等. 基于机器学习的低含油饱和度砂岩储层参数预测: 以准噶尔盆地夏子街油田夏77井区下克拉玛依组为例[J]. 石油实验地质, 2024, 46(5): 1123-1134.
	LIU Jun, ZHONG Jie, NI Zhen, et al. Machine learning⁃based prediction of low oil saturation sandstone reservoir parameters: A case study of Lower Karamay Formation in Xia 77 well block of Xiazijie Oilfield, Junggar Basin[J]. Petroleum Geology & Experiment, 2024, 46(5): 1123-1134.
[15]	武娟, 罗仁泽, 雷璨如, 等. 基于大语言模型的致密砂岩储层测井含水饱和度预测[J]. 天然气工业, 2024, 44(9): 77-87.
	WU Juan, LUO Renze, LEI Canru, et al. Prediction of water saturation in tight sandstone reservoirs from well log data based on the large language models(LLMs)[J]. Natural Gas Industry, 2024, 44(9): 77-87.
[16]	杨春生, 焦艳丽, 蔡东梅, 等. 井—震结合曲流型储层构型研究及应用: 以大庆长垣西部斜坡区为例[J]. 石油地球物理勘探, 2024, 59(6): 1330-1339.
	YANG Chunsheng, JIAO Yanli, CAI Dongmei, et al. Meandering reservoir architecture based on well-seismic analysis and its application:Taking the western slope area of Daqing Oilfield as an example[J]. Oil Geophysical Prospecting, 2024, 59(6): 1330-1339.
[17]	程敏华, 孟德伟, 王丽娟, 等. 致密砂岩气藏水平井差异化开发效果评价: 以鄂尔多斯盆地苏里格气田为例[J]. 中国矿业大学学报, 2023, 52(2): 354-363.
	CHENG Minhua, MENG Dewei, WANG Lijuan, et al. Evaluation of horizontal well differential development effect in tight sandstone gas reservoir: A case study of Sulige gas field, Ordos Basin[J]. Journal of China University of Mining & Technology, 2023, 52(2): 354-363.
[18]	冀光, 贾爱林, 孟德伟, 等. 大型致密砂岩气田有效开发与提高采收率技术对策: 以鄂尔多斯盆地苏里格气田为例[J]. 石油勘探与开发, 2019, 46(3): 602-612. doi: 10.11698/PED.2019.03.19
	JI Guang, JIA Ailin, MENG Dewei, et al. Technical strategies for effective development and gas recovery enhancement of a large tight gas field: A case study of Sulige gas field, Ordos Basin, NW China[J]. Petroleum Exploration and Development, 2019, 46(3): 602-612.
[19]	韩扬. 数据清洗和支持向量机技术研究[D]. 天津: 河北工业大学, 2015.
	HAN Yang. Research on technologies of data cleaningand support vector machine[D]. Tianjin: Hebei University of Technology, 2015.
[20]	张豹. 基于机器学习算法的数据分类与标准化方法研究[J]. 信息与电脑(理论版), 2024, 36(6): 53-55.
	ZHANG Bao. Research on data classification and standardization methods based on machine learning algorithms[J]. Information & Computer, 2024, 36(6): 53-55.
[21]	江丽, 张智谟, 王琦玮, 等. 基于不同机器学习模型的石油测井数据岩性分类对比研究[J]. 物探与化探, 2024, 48(2): 489-497.
	JIANG Li, ZHANG Zhimo, WANG Qiwei, et al. Comparative study on lithology classification of oil logging data based on different machine learning models[J]. Geophysical and Geochemical Exploration, 2024, 48(2): 489-497.
[22]	HUANG G B, CHEN L, SIEW C K. Universal approximation using incremental constructive feedforward networks with random hidden nodes[J]. IEEE Transactions on Neural Networks, 2006, 17(4): 879-892.
[23]	SHI S Z, SHI G F, PEI J B, et al. Porosity prediction in tight sandstone reservoirs based on a one–dimensional convolutional neural network–gated recurrent unit model[J]. Applied Geophysics, 2023.
[24]	ROSTAMI A, BAGHBAN A, MOHAMMADI A H, et al. Rigorous prognostication of permeability of heterogeneous carbonate oil reservoirs: Smart modeling and correlation development[J]. Fuel, 2019, 236: 110-123.
[25]	李宁, 徐彬森, 武宏亮, 等. 人工智能在测井地层评价中的应用现状及前景[J]. 石油学报, 2021, 42(4): 508-522. doi: 10.7623/syxb202104008
	LI Ning, XU Binsen, WU Hongliang, et al. Application status and prospects of artificial intelligence in well logging and formation evaluation[J]. Acta Petrolei Sinica, 2021, 42(4): 508-522. doi: 10.7623/syxb202104008
[26]	WU J, YIN X, XIAO H. Seeing permeability from images: Fast prediction with convolutional neural networks[J]. Science Bulletin, 2018, 63(18): 1215-1222. doi: 10.1016/j.scib.2018.08.006 pmid: 36751091
[27]	黎子豪, 蒋恕. 基于机器学习和SHAP算法的声波测井曲线重构及可解释性分析[J]. 地质科技通报, 2025, 44(1): 321-331.
	LI Zihao, JIANG Shu. Reconstructing and interpreting analysis of sonic logging curves based on machine learning and SHAP algorithm[J]. Bulletin of Geological Science and Technology, 2025, 44(1): 321-331.
[28]	田仁飞, 李山, 刘涛, 等. 基于XGBoost算法的v_P/v_S预测及其在储层检测中的应用[J]. 石油地球物理勘探, 2024, 59(4): 653-663.
	TIAN Renfei, LI Shan, LIU Tao, et al. v_P/v_S prediction based on XGBoost algorithm and its application in reservoir detection[J]. Oil Geophysical Prospecting, 2024, 59(4): 653-663.
[29]	张家臣, 邓金根, 谭强, 等. 基于XGBoost的测井曲线重构方法[J]. 石油地球物理勘探, 2022, 57(3): 697-705, 496.
	ZHANG Jiachen, DENG Jingen, TAN Qiang, et al. Reconstruction of well logs based on XGBoost[J]. Oil Geophysical Prospecting, 2022, 57(3): 697-705, 496.
[30]	FENG R, GRANA D, BALLING N. Imputation of missing well log data by random forest and its uncertainty analysis[J]. Computers & Geosciences, 2021, 152: 104763.
[31]	XIA J, ZHANG S, CAI G, et al. Adjusted weight voting algorithm for random forests in handling missing values[J]. Pattern Recognition, 2017, 69: 52-60.
[32]	韩如冰, 高严, 张元峰. 基于BP神经网络的分层相控碳酸盐岩储层渗透率预测方法[J]. 中国海上油气, 2024, 36(1): 100-108.
	HAN Rubing, GAO Yan, ZHANG Yuanfeng. Intelligent prediction method for permeability of layered phase controlled carbonate reservoirs based on BP neural network[J]. China Offshore Oil and Gas, 2024, 36(1): 100-108.
[33]	XIANG M, QIN P, ZHANG F. Research and application of logging lithology identification for igneous reservoirs based on deep learning[J]. Journal of Applied Geophysics, 2020, 173: 103929.
[34]	陈芊澍, 文晓涛, 何健, 等. 基于极限学习机的裂缝带预测[J]. 石油物探, 2021, 60(1): 149-156. doi: 10.3969/j.issn.1000-1441.2021.01.014
	CHEN Qianshu, WEN Xiaotao, HE Jian, et al. Prediction of a fracture zone using an extreme learning machine[J]. Geophysical Prospecting for Petroleum, 2021, 60(1): 149-156. doi: 10.3969/j.issn.1000-1441.2021.01.014
[35]	邓少贵, 张凤姣, 陈前, 等. 基于混合机器学习算法的页岩薄互层识别方法[J]. 石油学报, 2023, 44(7): 1097-1104. doi: 10.7623/syxb202307006
	DENG Shaogui, ZHANG Fengjiao, CHEN Qian, et al. Identification of shale thin interbeds based on hybrid machine learning algorithm[J]. Acta Petrolei Sinica, 2023, 44(7): 1097-1104. doi: 10.7623/syxb202307006
[36]	HUANG G B, ZHOU H, DING X, et al. Extreme learning machine for regression and multiclass classification[J]. IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), 2012, 42(2): 513-529.
[37]	高飞, 曲志鹏, 魏震, 等. 基于机器学习方法的测井岩相分类研究[J]. 地球物理学进展, 2024: 39(3): 1173-1192.
	GAO Fei, QU Zhipeng, WEI Zhen, et al. Study on well-log lithofacies classification based on machine learning methods[J]. Progress in Geophysics, 2024: 39(3): 1173-1192.
[38]	李盼池, 李文杰. 基于一维卷积神经网络的岩石物理相识别[J]. 吉林大学学报(信息科学版), 2022, 40(1): 51-63.
	LI Panchi, LI Wenjie. Identification of petrophysical facies based on one-dimensional convolutional neural networks[J]. Journal of Jilin University(Information Science Edition), 2022, 40(1): 51-63.
[39]	刘学锋, 张晓伟, 曾鑫, 等. 采用机器学习分割算法和扫描电镜分析页岩微观孔隙结构[J]. 中国石油大学学报(自然科学版), 2022, 46(1): 23-33.
	LIU Xuefeng, ZHANG Xiaowei, ZENG Xin, et al. Pore structure characterization of shales using SEM and machine learning-based segmentation method[J]. Journal of China University of Petroleum(Edition of Natural Science), 2022, 46(1): 23-33.
[40]	AO Y, ZHU L, GUO S, et al. Probabilistic logging lithology characterization with random forest probability estimation[J]. Computers & Geosciences, 2020, 144: 104556.
[41]	DURGA KANNAIAH P V, MAURYA N K. Machine learning approaches for formation matrix volume prediction from well logs: Insights and lessons learned[J]. Geoenergy Science and Engineering, 2023, 229: 212086.
[42]	王嵩然, 李潮流, 刘英明, 等. 高阻地层HDIL阵列感应影响因素分析与校正方法[J]. 石油科学通报, 2024, 9(1): 50-61.
	WANG Songran, LI Chaoliu, LIU Yingming, et al. Analysis of influence factors and method of correction in high resistivity formation by a high definition induction log[J]. Petroleum Science Bulletin, 2024, 9(1): 50-61.

层号	厚度/m	岩性	自然伽马/API	密度/（g/cm³）	深侧向电阻率/（Ω·m）	浅侧向电阻率/（Ω·m）
1	13.5	泥岩	122.3	2.52	17.5	22.1
2	5.5	砂岩	24.8	2.41	22.9	26.1
3	2.8	粉砂质泥岩	87.1	2.64	22.5	26.2
4	3.5	砂岩	31.6	2.48	25.6	28.3
5	2.2	泥岩	113.4	2.61	16.4	20.3

倾角θ/（°）	λ_S	λ_M	模型数量/个
70~75	1.00	1.2	50
75~80	1.00	1.3	50
80~85	1.00	1.4	50
85~90	1.00	1.5	70
70~75	1.05	1.2	50
75~80	1.05	1.3	50
80~85	1.05	1.4	50
85~90	1.05	1.5	70
70~75	1.10	1.2	50
75~80	1.10	1.3	50
80~85	1.10	1.4	50
85~90	1.10	1.5	70
70~75	1.15	1.2	50
75~80	1.15	1.3	50
80~85	1.15	1.4	50
85~90	1.15	1.5	70
70~75	1.20	1.2	50
75~80	1.20	1.3	50
80~85	1.20	1.4	50
85~90	1.20	1.5	70

算法名称	基本原理	优势	适用的目标变量
极端梯度提升树（eXtreme Gradient Boosting，XGBoost）	基于梯度提升决策树通过迭代添加决策树优化性能，目标函数包括损失函数和正则化项	快速的训练速度和较高的预测精度，良好的泛化能力	Y₂、Y₄、Y₅、Z₃、Z₄、Z₅
随机森林（Random Forest，RF）	集成多棵决策树提高模型的鲁棒性和泛化能力，每棵树在随机选择的样本和特征上训练，减少过拟合风险	有效减少过拟合，提高预测稳定性	Y₁、Y₂、Y₃、Y₄、Y₅、Z₁、Z₂、Z₃、Z₄、Z₅
支持向量机（Support Vector Machine，SVM）	在高维空间中构建最优超平面，通过优化数据点到超平面最小距离提高模型的鲁棒性	能处理高维数据，具有良好的泛化能力	不适用
轻量级梯度提升机（Light Gradient Boosting Machine，LightGBM）	通过对特征值进行直方图离散化提高训练速度和效率，同时保持高预测精度	训练速度快，能处理大规模数据集，预测精度高	Y₁、Y₃、Z₁、Z₂
分类提升机（Categorical Boosting，CatBoost）	优化的梯度提升算法，通过目标编码和其他优化技术提高对类别数据的处理能力	有效处理类别特征，提高预测精度和效率	Y₁、Y₂、Y₃、Y₄、Y₅、Z₁、Z₂、Z₃、Z₄、Z₅
极限学习机（Extreme Learning Machine，ELM）	通过随机生成的隐藏层提高训练速度和预测性能，无需迭代训练，具有很高的计算效率	训练速度快，大多数情况下预测精度高	不适用

目标变量	模型	MSE值	决定系数（R²）	MAE值
Y₁	RF	0.008 6	0.998 3	0.007 8
Y₂	RF	0.021 6	0.990 1	0.014 2
Y₃	RF	0.010 3	0.998 1	0.013 1
Y₄	RF	0.003 9	0.998 4	0.007 8
Y₅	XGBoost	0.000 7	0.997 4	0.005 3
Z₁	RF	0.024 3	0.998 6	0.097 0
Z₂	RF	0.000 7	0.999 8	0.097 0
Z₃	RF	0.000 5	0.998 8	0.009 9
Z₄	XGBoost	0.000 4	0.998 1	0.014 6
Z₅	RF	0.000 1	0.999 1	0.009 4

隐藏层节点数	激活函数	学习率	训练时间/s	验证集MSE值	验证集MAE值	验证集R²
16	ReLU	10^-4	115	0.020 5	0.120 1	0.995 0
32	ReLU	10^-4	132	0.016 0	0.105 0	0.997 8
64	ReLU	10^-4	204	0.008 2	0.007 1	0.999 1
128	ReLU	10^-4	310	0.007 7	0.006 8	0.999 2
64	Sigmoid	10^-4	220	0.010 3	0.011 4	0.996 5
64	Tanh	10^-4	215	0.016 5	0.010 8	0.997 2
64	ReLU	10^-5	250	0.009 1	0.008 3	0.999 5
64	ReLU	10^-3	150	0.013 4	0.115 0	0.995 2