基于地震属性约简的深度学习储层物性参数预测：以莺歌海盆地乐东区为例

刘仕友; 曲福良; 周凡; 邓利峰

doi:10.15953/j.ctta.2021.048

基于地震属性约简的深度学习储层物性参数预测：以莺歌海盆地乐东区为例

1.
中海石油（中国）有限公司海南分公司, 海口570000
2.
中国石油大学（华东）地球科学与技术学院, 山东青岛266580

基金项目: 中海油有限公司重大科技项目（南海西部油田上产2000万方关键技术研究（CNOOC-KJ 135 ZDXM38 ZJ02ZJ））。

详细信息

作者简介:
刘仕友: 男，中海石油（中国）有限公司海南分公司高级工程师，主要从事勘探地球物理方法应用研究，E-mail：liushiyou@139.com

曲福良: 男，中国石油大学（华东）地质资源与地质工程专业硕士研究生，主要从事储层地球物理方法与技术研究，E-mail：2411727425@qq.com

中图分类号: P 631.3；P 315；O 242
计量
- 文章访问数: 409
- HTML全文浏览量: 100
- PDF下载量: 61
出版历程
- 收稿日期: 2021-11-22
- 录用日期: 2022-02-25
- 网络出版日期: 2022-03-09
- 发布日期: 2022-09-30

Deep Learning Reservoir Parameter Prediction Based on Seismic Attribute Reduction: Take Ledong Area of Yinggehai Basin as an Example

1.
Haikou Branch, CNOOC China Limited, Haikou 570000, China
2.
School of Geosciences, China University of Petroleum (East China), Qingdao 266580, China

摘要

摘要: 储层物性参数作为描述储层特性、储层建模和流体模式的重要指标，其准确估算可以为储层预测提供有力参考依据，但传统储层物性参数反演方法无法兼顾反演精度及空间连续性。针对上述问题，本文引入地震属性作为深度学习算法输入，针对地震属性之间存在的信息冗余特征，利用随机森林-递归消除法对地震属性进行约简预处理，最终建立一种基于地震属性约简的储层物性参数预测方法。实际数据测试结果表明，地震属性约简的深度学习储层物性参数预测结果具有良好的精度及横向分辨率，证实本文方法的有效性。
- 深度学习 /
- 地震属性约简 /
- 随机森林算法 /
- 储层物性参数
Abstract: As an important indicator to describe reservoir characteristics, reservoir modeling and fluid model, the accurate estimation of reservoir physical parameters can provide a powerful reference for reservoir prediction, but the traditional inversion method of reservoir physical parameters can not give consideration to inversion accuracy and spatial continuity. To solve the above problems, this paper introduced seismic attributes as input of deep learning algorithm. Aiming at the information redundancy among seismic attributes, random forest-recursive elimination method was used to reduce the seismic attributes, thus a prediction method of reservoir physical property parameters based on seismic attribute reduction was finally established. The actual data test results showed that the prediction results of reservoir physical parameters by deep learning based on seismic attribute reduction presented good accuracy and lateral resolution, which confirmed the effectiveness of the proposed method.
- deep learning /
- seismic attribute reduction /
- random forest algorithm /
- reservoir physical parameters

HTML全文

图 1 深度神经网络结构示意图

Figure 1. Deep neural network structure diagram

下载: 全尺寸图片幻灯片

图 2 RF-RFE算法流程图

Figure 2. The RF-RFE algorithm flow chart

下载: 全尺寸图片幻灯片

图 3 地震属性重要度排序图

Figure 3. Importance ranking diagram of seismic attributes

下载: 全尺寸图片幻灯片

图 4 决定系数随地震属性个数变化图

Figure 4. Coefficient of determination varies with the number of seismic attributes

下载: 全尺寸图片幻灯片

图 5 训练集孔隙度真实值与预测值对比

Figure 5. Contrast the actual and predicted values of the porosity curve in train set

下载: 全尺寸图片幻灯片

图 6 测试集孔隙度真实值与预测值对比

Figure 6. Contrast the actual and predicted values of the porosity curve in test set

下载: 全尺寸图片幻灯片

图 7 孔隙度测井曲线与孔隙度反演剖面

Figure 7. Porosity logs and porosity inversion profiles

下载: 全尺寸图片幻灯片

图 8 含油气性测井解释结果与孔隙度反演剖面

Figure 8. Oil and gas logging interpretation and porosity inversion profile

下载: 全尺寸图片幻灯片

参考文献(26)

[1]	杨文采. 神经网络算法在地球物理反演中的应用[J]. 石油物探, 1995,34(2): 116−120. YANG W C. Application of neural network alogorithms to geophysical inversion[J]. Geophysical Prospecting for Petroleum, 1995, 34(2): 116−120. (in Chinese).
[2]	印兴耀, 吴国忱, 张洪宙. 神经网络在储层横向预测中的应用[J]. 石油大学学报(自然科学版), 1994,18(5): 20−26. YIN X Y, WU G C, ZHANG H Z. The application of neural networks in the reservior prediction[J]. Journal of China University of Petroleum (Edition of Natural Science), 1994, 18(5): 20−26. (in Chinese).
[3]	印兴耀, 杨风丽, 吴国忱. 神经网络在CB油田储层预测和储层厚度计算中的应用[J]. 石油大学学报(自然科学版), 1998,22(2): 3−5. YIN X Y, YANG F L, WU G C. Application of neural network to predicting reservoir and calculating thickness in CB oilfield[J]. Journal of China University of Petroleum (Edition of Natural Science), 1998, 22(2): 3−5. (in Chinese).
[4]	XU Y, YIN C, PAN Y, et al. First-break automatic picking technology based on semantic segmentation[J]. Geophysical Prospecting, 2021, 69(6): 1181-1207.
[5]	孙小东, 王伟奇, 任丽娟, 等. 地震数据智能去噪与传统去噪方法的对比及展望[J]. 地球物理学进展, 2022, 35(6): 2211-2219. SUN X D, WANG W Q, REN L J, et al. Comparison and prospect on AI denoising of seismic data along with traditional denoising methods[J]. Progress in Geophysics, 2020, 35(6): 2211-2219. (in Chinese).
[6]	桑凯恒, 张繁昌. 基于模糊粗糙集的机器学习储层参数预测[J]. CT理论与应用研究, 2018,27(4): 455−464. DOI: 10.15953/j.1004-4140.2018.27.04.05. SANG K H, ZHANG F C. Prediction of reservoir parameters of machine learning based on fuzzy rough set[J]. CT Theory and Applications, 2018, 27(4): 455−464. DOI: 10.15953/j.1004-4140.2018.27.04.05. (in Chinese).
[7]	WU X, LIANG L, SHI Y, et al. FaultSeg3D: Using synthetic data sets to train an end-to-end convolutional neural network for 3D seismic fault segmentation[J]. Geophysics, 2019, 84(3): IM35−IM45. doi: 10.1190/geo2018-0646.1
[8]	WU X, YAN S, QI J, et al. Deep learning for characterizing paleokarst collapse features in 3D seismic images[J]. Journal of Geophysical Research: Solid Earth, 2020.
[9]	安鹏, 曹丹平, 赵宝银, 等. 基于LSTM循环神经网络的储层物性参数预测方法研究[J]. 地球物理学进展, 2019, 34(5): 1849-1858. AN P, CAO D P, ZHAO B Y, et al. Reservoir physical parameters prediction based on LSTM recurrent neural network[J]. Progress in Geophysics, 2019, 34(5): 1849-1858. (in Chinese).
[10]	安鹏, 曹丹平. 基于深度学习的测井岩性识别方法研究与应用[J]. 地球物理学进展, 2018,33(3): 1029−1034. doi: 10.6038/pg2018BB0319 AN P, CAO D P. Research and application of logging lithology identification based on deep learning[J]. Progress in Geophysics, 2018, 33(3): 1029−1034. (in Chinese). doi: 10.6038/pg2018BB0319
[11]	闫星宇, 顾汉明, 罗红梅, 等. 基于改进深度学习方法的地震相智能识别[J]. 石油地球物理勘探, 2020,55(6): 1169−1177. YAN X Y, GU H M, LUO H M, et al. Intelligent seismic facies classification based on an improved deep learning method[J]. Oil Geophysical Prospecting, 2020, 55(6): 1169−1177. (in Chinese).
[12]	陈康, 狄贵东, 张佳佳, 等. 基于改进U-Net卷积神经网络的储层预测[J]. CT理论与应用研究, 2021,30(4): 403−416. DOI: 10.15953/j.1004-4140.2021.30.04.01. CHEN K, DI G D, ZHANG J J, et al. Reservoir prediction based on improved U-Net convolutional neural network[J]. CT Theory and Applications, 2021, 30(4): 403−416. DOI: 10.15953/j.1004-4140.2021.30.04.01. (in Chinese).
[13]	PURVES S, ALAEI B, BONAS J B, et al. Machine learning from core to seismic; Porosity prediction in the Norwegian North Sea[C]//81st EAGE Conference and Exhibition 2019. European Association of Geoscientists & Engineers, 2019, 2019(1): 1-5.
[14]	印兴耀, 周静毅. 地震属性优化方法综述[J]. 石油地球物理勘探, 2005,40(4): 482−489. doi: 10.3321/j.issn:1000-7210.2005.04.027 YIN X Y, ZHOU J Y. Summary of optimum methods of seismic attributes[J]. Oil Geophysical Prospecting, 2005, 40(4): 482−489. (in Chinese). doi: 10.3321/j.issn:1000-7210.2005.04.027
[15]	郭淑文, 程然, 祝文亮, 等. 数据挖掘技术在地震属性降维中的应用[J]. 天然气地球科学, 2010,21(4): 670−677. GUO S W, CHENG R, ZHU W L, et al. Application of data mining technique in reduction of dimensions of seismic attribute parameters[J]. Natural Gas Geoscience, 2010, 21(4): 670−677. (in Chinese).
[16]	ROWEIS S, SAUL L. Nonlinear dimensionality reduction by locally linear embedding[J]. Science, 2000, 290(5500): 2323−2326. doi: 10.1126/science.290.5500.2323
[17]	CHEN Q, SIDNEY S. Advances in seismic attribute technology[J]. Seg Technical Program Expanded Abstracts, 1949, 16(1): 2067.
[18]	王全才. 随机森林特征选择[D]. 大连: 大连理工大学, 2011. WANG Q C. Random forest feature selection[D]. Dalian: Dalian University of Technology, 2011. (in Chinese).
[19]	姚登举, 杨静, 詹晓娟. 基于随机森林的特征选择算法[J]. 吉林大学学报(工学版), 2014(1): 142-146. YAO D J, YANG J, ZHAN X J. Feature selection algorithm based on random forest[J]. Journal of Jilin University (Engineering and Technology Edition), 2014(1): 142-146. (in Chinese).
[20]	宁永鹏. 高维小样本数据的特征选择研究及其稳定性分析[D]. 厦门: 厦门大学, 2014. NING Y P. Study on feature selection and stability analysis of high dimensional small sample data[D]. Xiamen: Xiamen University, 2014.
[21]	XIE S, GIRSHICK R, DOLLÁR P, et al. Aggregated residual transformations for deep neural networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017: 1492-1500.
[22]	SHIRO T, YUKI Y, MASATO O. Impact of layer normalization on single-layer perceptron: Statistical mechanical analysis[J]. Journal of the Physical Society of Japan, 2019, 88(7).
[23]	SHOUJI S, SHIGEKO S I, YOUICHI K. Normalization processes an topographic mapping model between cell layers represented by undirected graphs[J]. Journal of Japan Society for Fuzzy Theory and Systems, 2017, 14(1): 43-54.
[24]	董师师, 黄哲学. 随机森林理论浅析[J]. 集成技术, 2013,2(1): 1−7. doi: 10.5121/ijite.2013.2101 HUANG S S, HUANG Z X. A brief theoretical overview of random forests[J]. Journal of Integration Technology, 2013, 2(1): 1−7. (in Chinese). doi: 10.5121/ijite.2013.2101
[25]	周雪晴, 张占松, 张超谟, 等. 基于粗糙集-随机森林算法的复杂岩性识别[J]. 大庆石油地质与开发, 2017,(6): 131−137. ZHOU X Q, ZHANG Z S, ZHANG C M. et al. Complex lithologic iidentification based on rough set-random forest algorism[J]. Petroleum Geology & Oilfield Development in Daqing, 2017, (6): 131−137. (in Chinese).
[26]	胥雪炎, 李补喜. 不同被解释变量选择对决定系数R²的影响研究[J]. 太原科技大学学报, 2007,28(5): 363−365. XU X Y, LI B X. Research on the effect of selection of dependent variables on R² statistic[J]. Journal of Taiyuan University of Science and Technology, 2007, 28(5): 363−365. (in Chinese).