基于深度生成模型的河道砂建模方法

高小洋; 何文祥; 胡勇

doi:10.19509/j.cnki.dzkq.2022.0208

基于深度生成模型的河道砂建模方法

doi: 10.19509/j.cnki.dzkq.2022.0208

高小洋^{a, b,},
何文祥^{a, b},
胡勇^{a, b, ,}

a.
长江大学资源与环境学院, 武汉 430100
b.
长江大学油气地球化学与环境湖北省重点实验室, 武汉 430100

详细信息

作者简介:
高小洋(1995-), 男, 现正攻读矿产普查与勘探专业博士学位, 主要从事地球化学和人工智能地质建模研究工作。E-mail: 18734311896@163.com

通讯作者:
胡勇(1980-), 男, 副教授, 主要从事油藏沉积学和地质建模研究工作。E-mail: 64421847@qq.com

中图分类号: P628
计量
- 文章访问数: 488
- PDF下载量: 54
- 被引次数: 0
出版历程
- 收稿日期: 2021-09-01

A modeling method of channel sand based on deep generation models

Gao Xiaoyang^{a, b
,},
He Wenxiang^{a, b},
Hu Yong^{a, b
, ,}

a.
College of Resources and Environment, Yangtze University, Wuhan 430100, China
b.
Hubei Key Laboratory of Petroleum Geochemistry and Environment, Yangtze University, Wuhan 430100, China

摘要

摘要:
在传统的河道砂体建模方法中，基于目标建模的方法难以条件化，多点地质统计学难以再现连续河道砂体的形态，导致建模成果难以直接应用于油田生产。深度生成模型通过深度学习，可以生成足够精确的河道砂体模型，能再现复杂的河道砂体形态，很好地满足井点条件，弥补了传统建模算法的不足。在建模过程中，首先基于目标模拟方法与计算机匹配操作建立了20 000个河道砂体模型与对应的条件集，并结合变分自编码（VAE）与生成对抗网络（GAN）的理论，建立深度生成模型，其中包括分类器、编码器、解码器与判别器。将条件数据与真实模型输入深度生成模型中得到对应的河道砂体模型，通过大量的训练建立了可以生成满足井点条件的河道砂体的生成器，最后将井点数据输入生成器中建立相应的河道砂体模型。研究结果表明，深度生成模型建模算法与传统建模算法相比不仅展现出了连续、清晰的河道砂体，并且在给定的条件下可以生成多个河道砂体模型。该建模方法克服了传统河道砂建模方法的不足，为河道砂体储层建模提供了新的解决思路，建立的河道模型可为油田开发提供参考。
- 人工智能 /
- 深度生成模型 /
- 储层建模 /
- 河道砂体
Abstract:
In traditional channel sand body modeling methods, the method based on target modeling is difficult to condition, and multipoint geostatistics have difficulty reproducing the shape of continuous channel sand bodies, which makes the modeling results difficult to directly apply to oilfield production. Through depth learning, the depth generation model can generate a sufficiently accurate channel sand body model, reproduce the complex channel sand body shape, meet the well point conditions, and compensate for the shortcomings of the traditional modeling algorithm. In the process of modeling, 20 000 channel sand body models and corresponding condition sets are established based on the target simulation method and computer matching operation. Combined with the theory of variational autoencoders (VAE) and generative adversarial networks (GAN), a depth generation model is established, including a classifier, encoder, decoder and discriminator. The condition data and the real model are input into the depth generation model to obtain the corresponding channel sand body model. Through extensive training, a generator that can generate channel sand bodies that meet the well point conditions is established. Finally, the well point data are input into the generator to establish the corresponding channel sand body model. The results show that compared with the traditional modeling algorithm, the depth generation model modeling algorithm not only shows a continuous and clear channel sand body but can also generate multiple channel sand body models under given conditions. This modeling method overcomes the shortcomings of traditional channel sand modeling methods and provides a new solution for channel sand reservoir modeling. The established channel model can provide a reference for the oilfield development stage.
- artificial intelligence /
- deep generation model /
- reservoir modeling /
- channel sand body

HTML全文

图 1 VAE结构图(实线表示生成模型，虚线表示变分近似)

z.隐变量；x.可观测量；Ф.变分近似的相关参数；θ.生成模型的相关参数；N.样本数

Figure 1. VAE structural diagram

下载: 全尺寸图片幻灯片

图 2 GAN结构图

Figure 2. GAN structural diagram

下载: 全尺寸图片幻灯片

图 3 深度生成模型训练过程

Figure 3. Training process of the deep generative model

下载: 全尺寸图片幻灯片

图 4 随机生成的河道砂体模型数据集(此处只显示了16个模型的数据)

Figure 4. Randomly generated data set of channel sand body model

下载: 全尺寸图片幻灯片

图 5 输入训练的条件集

Figure 5. Condition set of input training

下载: 全尺寸图片幻灯片

图 6 用于测试的井点图

Figure 6. Well point diagram for testing

下载: 全尺寸图片幻灯片

图 7 条件数据与河道模型通过one_hot特征进行连接

Figure 7. Conditional data and channel model are connected through one_hot features

下载: 全尺寸图片幻灯片

图 8 编码器E与解码器G的结构图

Figure 8. Structural diagram of Encoder E and Decoder G

下载: 全尺寸图片幻灯片

图 9 分类器C1、C2损失函数变化曲线

Figure 9. Variation curve of loss function of the classifier C1, C2

下载: 全尺寸图片幻灯片

图 10 深度生成模型损失函数

Figure 10. Loss function of the deep generative model

下载: 全尺寸图片幻灯片

图 11 训练精度(a)和测试精度(b)变化曲线

Figure 11. Change curve of training accuracy (a) and testing accuracy (b)

下载: 全尺寸图片幻灯片

图 12 生成的模型

Figure 12. Generated model

下载: 全尺寸图片幻灯片

图 13 训练不同的迭代次数生成的河道砂体模型

Figure 13. Training different numbers of epochs to generate the channel sand body model

下载: 全尺寸图片幻灯片

图 14 VAE生成的河道砂体模型

Figure 14. Channel sand body model generated by the VAE

下载: 全尺寸图片幻灯片

图 15 基于目标模拟生成的河道砂体模型

Figure 15. Channel model generated based on target simulation

下载: 全尺寸图片幻灯片

图 16 多点统计生成的河道砂体模型

Figure 16. Channel sand body model generated by multipoint statistics

下载: 全尺寸图片幻灯片

参考文献(35)

[1]	束青林. 河道砂储层油藏动态模型和剩余油预测[M]. 北京: 石油工业出版社, 2004. Shu Q L. Reservoir performance model and remaining oil prediction of channel sand reservoir[M]. Beijing: Petroleum Industry Press, 2004(in Chinese).
[2]	李勤, 张利锋, 孙丽. 遗传模拟退火算法在储层属性建模中的应用[J]. 大庆石油地质与开发, 2005(1): 31-32, 106. doi: 10.3969/j.issn.1000-3754.2005.01.009 Li Q, Zhang L F, Sun L. Influence of asphaltene deposition on oil seepage characteristics[J]. Petroleum Geology & Oilfield Development in Daqing, 2005(1): 31-32, 106(in Chinese with English abstract). doi: 10.3969/j.issn.1000-3754.2005.01.009
[3]	陈建阳, 于兴河, 张志杰, 等. 储层地质建模在油藏描述中的应用[J]. 大庆石油地质与开发, 2005(3): 17-18, 104. doi: 10.3969/j.issn.1000-3754.2005.03.006 Chen J Y, Yu X H, Zhang Z J, et al. Application of reservoir modeling in reservoir description of Baolige Oil Field[J]. Petroleum Geology & Oilfield Development in Daqing, 2005(3): 17-18, 104(in Chinese with English abstract). doi: 10.3969/j.issn.1000-3754.2005.03.006
[4]	徐伟, 房磊, 刘钧, 等. 浅层高渗砂岩油藏分级相控地质建模[J]. 地质科技情报, 2017, 36(1): 197-201. https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ201701024.htm Xu W, Fang L, Liu J, et al. Hierarchic facies-constrained geological modeling of shallow high permeability sandy reservoir[J]. Geological Science and Technology Information, 2017, 36(1): 197-201(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ201701024.htm
[5]	付志国, 石成方, 赵翰卿, 等. 喇萨杏油田河道砂岩厚油层夹层分布特征[J]. 大庆石油地质与开发, 2007(4): 55-58. doi: 10.3969/j.issn.1000-3754.2007.04.013 Fu Z G, Shi C F, Zhao H Q, et al. The distribution characteristics of interlayer in thick channel sand oil reservoir in Lasaxing Oilfield[J]. Petroleum Geology & Oilfield Development in Daqing, 2007(4): 55-58(in Chinese with English abstract). doi: 10.3969/j.issn.1000-3754.2007.04.013
[6]	舒志华, 张立有, 刘刚. 复合砂体中单一河道的识别方法[J]. 大庆石油地质与开发, 2006(4): 18-20, 120. doi: 10.3969/j.issn.1000-3754.2006.04.007 Shu Z H, Zhang L Y, Liu G. Identification of single channel in compound sand body[J]. Petroleum Geology & Oilfield Development in Daqing, 2006(4): 18-20, 120(in Chinese with English abstract). doi: 10.3969/j.issn.1000-3754.2006.04.007
[7]	赵翰卿. 高分辨率层序地层对比与我国的小层对比[J]. 大庆石油地质与开发, 2005(1): 5-9, 12-105. doi: 10.3969/j.issn.1000-3754.2005.01.002 Zhao H Q. High-resolution sequential stratigraphy correlation and Chinese subzone correlation[J]. Petroleum Geology & Oilfield Development in Daqing, 2005(1): 5-9, 12-105(in Chinese with English abstract). doi: 10.3969/j.issn.1000-3754.2005.01.002
[8]	何宇航, 于开春. 分流平原相复合砂体单一河道识别及效果分析[J]. 大庆石油地质与开发, 2005(2): 17-19, 104. doi: 10.3969/j.issn.1000-3754.2005.02.006 He Y H, Yu K C. Single channel identification and effect analysis of compound sand body in distributary plain facies[J]. Petroleum Geology & Oilfield Development in Daqing, 2005(2): 17-19, 104(in Chinese with English abstract). doi: 10.3969/j.issn.1000-3754.2005.02.006
[9]	沈华, 尹微, 徐佑平. 提高砂岩油藏储层预测精度的方法[J]. 大庆石油地质与开发, 2005(3): 24-27, 104. doi: 10.3969/j.issn.1000-3754.2005.03.009 Shen H, Yin W, Xu Y P. Improving predicting accuracy of sandstone oil reservoirs[J]. Petroleum Geology & Oilfield Development in Daqing, 2005(3): 24-27, 104(in Chinese with English abstract). doi: 10.3969/j.issn.1000-3754.2005.03.009
[10]	Haldorsen H H, Chang D M. Notes on stochastic shales from outcrop to simulation models[C]//Lake L W, Carol H B. Reservoir characterization. New York, USA: Academic Press, 1986: 152-167.
[11]	Haldorsen H H, Damsleth E. Stochastic modelling[J]. J. Pet. Technol., 1990, 42(4): 404-412. doi: 10.2118/20321-PA
[12]	Holden L, Hauge R, Skare O, et al. Modeling of fluvial reservoirs with object models[J]. Math. Geol., 1998, 30: 473-496.
[13]	Strebelle S B, Journel A G. Reservoir modeling using multiple-point statistics[C]//Anon. SPE Paper 71324 Presented at the SPE Annual Technical Conference and Exhibition Held in New Orleans, Louisiana, 2001: 1-11.
[14]	Strebelle S B. Conditional simulation of complex geological structures using multiple-point statistics[J]. Math. Geol., 2002, 34: 1-21. doi: 10.1023/A:1014009426274
[15]	Mariethoz G, Renard P, Straubhaar J. The direct sampling method to perform multiple-point geostatistical simulations[J]. Water Resour. Res., 2010, 46: 11536
[16]	Rezaee H, Marcotte D, Tahmasebi P, et al. Multiple-point geostatistical simulation using enriched pattern databases[J]. Stoch. Environ. Res. Risk Assess., 2015, 29: 893-913. doi: 10.1007/s00477-014-0964-6
[17]	陈麒玉, 刘刚, 何珍文, 等. 面向地质大数据的结构-属性一体化三维地质建模技术现状与展望[J]. 地质科技通报, 2020, 39(4): 51-58. https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ202004007.htm Chen Q Y, Liu G, He Z W, et al. Current situation and prospect of structure-attribute integrated 3D geological modeling technology for geological big data[J]. Bulletin of Geological Science and Technology, 2020, 39(4): 51-58(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ202004007.htm
[18]	吴胜和, 李文克. 多点地质统计学: 理论、应用与展望[J]. 古地理学报, 2005(1): 137-144. https://www.cnki.com.cn/Article/CJFDTOTAL-GDLX200501014.htm Wu S H, Li W K. Multiple-point geostatistics: Theory, application and perspective[J]. Journal of Palaeogeography, 2005(1): 137-144(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-GDLX200501014.htm
[19]	郝慧珍, 顾庆, 胡修棉. 基于机器学习的矿物智能识别方法研究进展与展望[J]. 地球科学, 2021, 46(9): 3091-3106. https://www.cnki.com.cn/Article/CJFDTOTAL-DQKX202109005.htm Hao H Z, Gu Q, Hu X M. Research advances and prospective in mineral intelligent identification based on machine learning[J]. Earth Science, 2021, 46(9): 3091-3106(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-DQKX202109005.htm
[20]	郭艳军, 周哲, 林贺洵, 等. 基于深度学习的智能矿物识别方法研究[J]. 地学前缘, 2020, 27(5): 39-47. https://www.cnki.com.cn/Article/CJFDTOTAL-DXQY202005006.htm Guo Y J, Zhou Z, Lin H X, et al. The mineral intelligence identification method based on deep learning algorithms[J]. Earth Science Frontiers, 2020, 27(5): 39-47(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-DXQY202005006.htm
[21]	刘彦锋, 张文彪, 段太忠, 等. 深度学习油气藏地质建模研究进展[J]. 地质科技通报, 2021, 40(4): 235-241. https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ202104023.htm Liu Y F, Zhang W B, Duan T Z, et al. Progress of deep learning in oil and gas reservoir geological modeling[J]. Bulletin of Geological Science and Technology, 2021, 40(4): 235-241(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ202104023.htm
[22]	马瑶, 赵江南. 机器学习方法在矿产资源定量预测应用研究进展[J]. 地质科技通报, 2021, 40(1): 132-141. https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ202101013.htm Ma Y, Zhao J N. Advances in the application of machine learning methods in mineral prospectivity mapping[J]. Bulletin of Geological Science and Technology, 2021, 40(1): 132-141(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ202101013.htm
[23]	万里明, 吴均, 卢军凯, 等. 基于Adam-神经网络的致密砂岩脆性评价方法: 以南堡凹陷高北边坡为例[J]. 地质科技通报, 2020, 39(2): 94-103. https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ202002011.htm Wan L M, Wu J, Lu J K, et al. Brittleness evaluation method of tight sandstone based on Adam-neural network: A case study of a block in Gaobei slope, Nanpu Sag[J]. Bulletin of Geological Science and Technology, 2020, 39(2): 94-103(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ202002011.htm
[24]	陈仲为. 深度学习的发展以及应用[J]. 现代计算机, 2019(17): 46-50. https://www.cnki.com.cn/Article/CJFDTOTAL-XDJS201917011.htm Chen Z W. Development and application of deep learning[J]. Modern Computer, 2019(17): 46-50(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-XDJS201917011.htm
[25]	张福临. 深度学习在岩石薄片图像生成中的研究与应用[D]. 西安: 西安石油大学, 2021. Zhang F L. Research and application of depth learning in rock slice image generation[D]. Xi'an: Xi'an Shiyou University, 2021(in Chinese with English abstract).
[26]	李兴保. 基于生成对抗网络的数字岩心重构研究[D]. 合肥: 合肥工业大学, 2020. Li X B. Research on digital core reconstruction based on generative countermeasure network[D]. Hefei: Hefei University of Technology, 2020(in Chinese with English abstract).
[27]	张挺, 王先武, 杜奕, 等. 基于DCGANs的二维页岩图像重构方法[J]. 上海电力大学学报, 2021, 37(4): 402-406. https://www.cnki.com.cn/Article/CJFDTOTAL-DYXY202104016.htm Zhang T, Wang X W, Du Y, et al. 2D shale image reconstruction based on DCGANs[J]. Journal of Shanghai University of Electric Power, 2021, 37(4): 402-406(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-DYXY202104016.htm
[28]	陈龙. 基于生成对抗网络的多孔介质重构[D]. 西安: 长安大学, 2020. Chen L. Reconstruction of porous media based on generative countermeasure network[D]. Xi'an: Chang'an University, 2020(in Chinese with English abstract).
[29]	Diederik P. Kingma and max welling, auto-encoding variational bayes[J]. CoRR, 2013, abs/1312.6114.
[30]	Goodfellow I, Pouget-Abadiej, Mirza M, et al. Generative adversarial nets[C]//Anon. Proceedings of the 2014 Conference on Advances in Neural Information Processing Systems, [S. l. ]: [s. n. ], 2014: 2672-2680.
[31]	Canchumuni S W A, Emerick A A, Pacheco M A C. Towards a robust parameterization for conditioning facies models using deep variational autoencoders and ensemble smoother[J]. Computers and Geosciences, 2019, 128: 87-102.
[32]	Laloy E, Héraut R, Jacques D, et al. Training-image based geostatistical inversion using a spatial generative adversarial neural network[J]. Water Resources Research, 2018, 54(1): 381-406.
[33]	Zhang T F, Peter T K, Emilien D, et al. Generating geologically realistic 3D reservoir facies models using deep learning of sedimentary architecture with generative adversarial networks[J]. Petroleum Science, 2019, 16: 541-549.
[34]	Bao J M, Chen D, Wen F, et al. CVAE-GAN: Fine-grained image generation through asymmetric training[C]//Anon. Proceedings of the IEEE International Conference on Computer Vision. [S. l. ]: [s. n. ], 2017: 2745-2754.
[35]	Kingma D, Ba J. Adam: A method for stochastic optimization[C]//Anon. Proceedings of International Conference on Learning Representations. [S. l. ]: [s. n. ], 2015.