基于随机森林算法的油气层敏感性损害预测

盛科鸣; 蒋官澄

doi:10.12358/j.issn.1001-5620.2023.04.002

基于随机森林算法的油气层敏感性损害预测

doi: 10.12358/j.issn.1001-5620.2023.04.002

盛科鸣^1,,
蒋官澄^2, ,

1.
中国石油大学(北京)人工智能学院, 北京 102249
2.
中国石油大学(北京)油气资源与探测国家重点实验室·中国石油大学(北京)石油工程教育部重点实验室·中国石油大学(北京), 北京 102249

基金项目: 国家自然科学基金青年科学基金项目“智能钻井液聚合物处理剂刺激响应机理与分子结构设计方法研究”（52004297）；中国博士后创新人才支持计划“大温差智能响应机理及智能恒流变无土相生物油基钻井液研究”（BX20200384）。

详细信息

作者简介:
盛科鸣，在读博士研究生，1997年生，研究方向为油气工程信息化与智能化技术。E-mail: keming@student.cup.edu.cn。

通讯作者:
蒋官澄，博士，二级教授，1966年生，研究领域为油田化学、储层保护等。E-mail: jgc5786@126.com。

中图分类号: TE258
计量
- 文章访问数: 465
- HTML全文浏览量: 189
- PDF下载量: 108
- 被引次数: 0
出版历程
- 收稿日期: 2023-02-07
- 修回日期: 2023-03-14
- 刊出日期: 2023-07-30

Prediction of Four Kinds of Sensibility Damages to Hydrocarbon Reservoirs Based on Random Forest Algorithm

SHENG Keming^1
,,
JIANG Guancheng^{2
, ,}

1.
College of Artificial Intelligence, China University of Petroleum (Beijing), Beijing 102249
2.
State Key Laboratory of Petroleum Resources and Prospecting·MOE Key Laboratory of Petroleum Engineering·China University of Petroleum (Beijing), Beijing 102249

摘要

摘要: 储层损害贯穿在油气田勘探开发的各个时期，其种类繁多、损害机理十分复杂。传统岩心流动实验评价储层敏感性的结果可靠，但岩心获取成本高、投入时间和成本大。调研和实践表明，利用神经网络、随机森林等算法基于小规模样本建立的模型可以实现对样本的预测，节约时间和经济成本。基于X区块敏感性室内评价小规模样本资料，选择训练集及测试集，深入对比了BP神经网络算法、径向基函数神经网络算法、随机森林算法，优选出随机森林算法作为储层敏感性损害定量诊断的主要方法，采用网格搜索等算法进行了超参数优化、根据因素权重对数据进行降维，以此提高预测精度，搭建了完整的模型。4种损害模型的R²平均值为0.852，预测精度在90.00%～95.68%。
- 储层敏感性预测 /
- 油气人工智能 /
- 随机森林 /
- 神经网络 /
- 相关性分析
Abstract: Many kinds of hydrocarbon reservoir damages with complex mechanisms have been encountered in every phase of oil and gas field exploration and development. Conventional core flow test used in evaluating the sensibility damage of a reservoir can give reliable test results, however, this test is both expensive (coring, for instance) and time consuming. Researches have shown that a model established with neural network and random forest algorithm on small-scale samples can be used to save time and money in predicting the properties of samples. In this study, the data of a set of small-scale samples tested in laboratory is obtained from the block X. The training-sets and testing-sets are then selected on the samples. By extensively comparing the results of three algorithms, which are the BP neural network algorithm, the radial basis function neural network algorithm and the random forest algorithm, the random forest algorithm is finally selected as the main method of quantitatively diagnosing the sensitivity damage of hydrocarbon reservoirs. To improve the prediction accuracy, algorithms such as grid search are used in hyperparameter optimization, and data dimensionality reduction is performed based on factor weight. A complete model is finally established based on the studies conducted. The average R² value of the four kinds of reservoir damage model is 0.852, with a prediction accuracy between 90.00% and 95.68%.
- Reservoir sensibility prediction /
- Oil and gas AI /
- Radom forest /
- Neural network /
- Correlation analysis

HTML全文

图 1 随机森林网络结构

下载: 全尺寸图片幻灯片

图 2 速敏Pearson相关矩阵

下载: 全尺寸图片幻灯片

图 3 速敏Spearman相关矩阵

下载: 全尺寸图片幻灯片

图 4 4种方法下BP神经网络迭代次数

下载: 全尺寸图片幻灯片

图 5 4种模型的RBF神经网络迭代次数

下载: 全尺寸图片幻灯片

图 6 网格搜索结果

下载: 全尺寸图片幻灯片

图 7 速敏模型因素权重

下载: 全尺寸图片幻灯片

图 8 3种算法的RMSE

下载: 全尺寸图片幻灯片

图 9 3种算法的R²平均大小

下载: 全尺寸图片幻灯片

表 1 4类BP神经网络的神经元个数

神经网络层	速敏	水敏	酸敏	碱敏
输入层	14	14	14	12
隐藏层	24	24	23	23
输出层	2	2	1	2

下载: 导出CSV

表 2 训练集与测试集的划分

敏感性类型	速敏	水敏	酸敏	碱敏
训练集	111	156	163	148
测试集	11	15	16	14

下载: 导出CSV

表 3 速敏预测结果

序号	速敏指数			临界流速
序号	真实值	预测值	误差/%	真实值	预测值	误差/%
1	43.24	43.9143	1.56	0.50	0.5304	6.08
2	23.3	23.2387	0.26	5.00	5.1206	2.41
3	27.0	26.7470	0.94	0.50	0.5213	4.26
4	35.0	38.8436	10.98	0.50	0.5619	12.38
5	54.0	49.7589	7.85	0.25	0.2703	8.12

下载: 导出CSV

表 4 预测结果

损害类型	MSE/%	精确度/%
速敏	4.318	95.68
水敏	10.004	90.00
酸敏	6.984	93.02
碱敏	6.008	94.00

下载: 导出CSV

参考文献(24)

[1]	谢金秀. 储层敏感性实验与机制研究[J]. 石化技术,2021,28(1):139-140. XIE Jinxiu. Reservoir sensitivity experiment and mechanism research[J]. Petrochemical Technology, 2021, 28(1):139-140.
[2]	肖玉茹,何峰煜. 塔里木盆地北部西达里亚油气田三叠系储层敏感性评价[J]. 新疆地质,1996(4):364-374. XIAO Yuru, HE Fengyu. Appraisal of sensitivity of triassic reservior of west Daliya oil-gas field in northern Tarim basin[J]. Xinjiang Geology, 1996(4):364-374.
[3]	孙建孟,李召成,谭未一. 用单相关分析法快速预测储层敏感性[J]. 钻井液与完井液,1999(1):4-8. SUN Jianmeng, LI Zhaocheng, TAN Weiyi. Rapidly predicating the formation-sensitivity with analysing method of single correlation[J]. Drilling Fluid & Completion Fluid, 1999(1):4-8.
[4]	王峰,向祖平,陈中华,等. 利用响应曲面法校正气藏储层应力敏感性曲线[J]. 西南石油大学学报(自然科学版),2010,32(5):96-99,190. WANG Feng, XIANG Zuping, CHEN Zhonghua, et al. Correction of stress sensitivity curve of gas reservoir by response surface method[J]. Journal of Southwest Petroleum University(Science & Technology Edition), 2010, 32(5):96-99,190.
[5]	LIU Y L, RUI Z H. A storage-driven CO₂ EOR for a net-zero emission target[J]. Engineering, 2022, 18: 79-87.
[6]	MUNGAN N. Permeability reduction through changes in pH and salinity[J]. Journal of Petroleum Technology, 1965, 17(12):1449-1453.
[7]	BRYANT S L, BULLER D C. Formation damage from acid treatments[J]. SPE Production Engineering, 1990, 5(4):455-460. doi: 10.2118/17597-PA
[8]	彭春耀,鄢捷年,李玉凤. 预测储层潜在敏感性损害的新方法[J]. 钻井液与完井液,1999(2):4-10. PENG Chunyao, YAN Jienian, LI Yufeng. New ways of predicting reservoir damage of potential sensitivity[J]. Drilling Fluid & Completion Fluid, 1999(2):4-10.
[9]	张玄奇. 储层敏感性的灰色评价[J]. 大庆石油地质与开发,2004,23(6):60-62. ZHANG Xuanqi. Grey evaluation of reservoir sensitivity[J]. Petroleum Geology & Oilfield Development in Daqing, 2004, 23(6):60-62.
[10]	AL-MUDHAFAR W J, RAO D N, SRINIVASAN S. Reservoir sensitivity analysis for heterogeneity and anisotropy effects quantification through the cyclic CO₂-assisted gravity drainage EOR process-a case study from south Rumaila oil field[J]. Fuel, 2018, 221:455-468.
[11]	ALEGRE L. An investigation of the applicability of expert systems to diagnose formation damage problems[J]. Los Angeles: University of Southern California, 1988.
[12]	郭建明,李棋,薄春生,等. 保护储集层综合集成决策支持系统的研制与应用[J]. 石油工业计算机应用,1995(3):22-28,32. GUO Jianming, LI Qi, BO Chunsheng, et al. Development and application of integrated decision support system for reservoir protection[J]. Computer Applications of Petroleum, 1995(3):22-28,32.
[13]	梅文荣,张绍槐. 基于神经网络的地层损害识别研究[J]. 西安石油学院学报(自然科学版),1995(1):46-49. MEI Wenrong, ZHANG Shaohuai. The research on the recognition of formation damage based on artificial neural network[J]. Journal of Xi'an Shiyou University (Natural Science Edition), 1995(1):46-49.
[14]	刘宝锋. 基于人工神经网络的超深井储层敏感性预测[D]. 青岛:中国石油大学(华东),2009. LIU Baofeng. The sensitivity prediction of the ultra-deep reservior based on artificial neural network[D]. Qingdao: China University of Petroleum (East China), 2009.
[15]	黄春,蒋官澄,纪朝凤,等. 基于径向基函数(RBF)神经网络的储层损害诊断技术研究[J]. 应用基础与工程科学学报,2010,18(2):313-320. HUANG Chun, JIANG Guancheng, JI Chaofeng, et al. Research on the formation damage diagnosis based on radial basis functions neural network[J]. Journal of Basic Science and Engineering, 2010, 18(2):313-320.
[16]	蒋官澄,王晓军,吴雄军,等. 模式识别在储层敏感性预测中的应用[J]. 油气地质与采收率,2010,17(5):61-64. JIANG Guancheng, WANG Xiaojun, WU Xiongjun, et al. Application of pattern recognition in the prediction of reservoir sensitivity[J]. Petroleum Geology and Recovery Efficiency, 2010, 17(5):61-64.
[17]	孙玉学,谢建波,才庆. 应用量子神经网络预测低渗储层水锁损害[J]. 特种油气藏,2012,19(6):53-55. SUN Yuxue, XIE Jianbo, CAI Qing. Prediction of water lock damage in low permeability reservoirs using quantum neural network[J]. Special Oil & Gas Reservoirs, 2012, 19(6):53-55.
[18]	高磊,潘树林. 基于遗传神经网络的储层敏感性预测方法研究[J]. 物探化探计算技术,2012,34(4):486-489. GAO Lei, PAN Shulin. Study on reservoir sensitivity prediction method based on genetic neural network[J]. Computing Techniques for Geophysical and Geochemical Exploration, 2012, 34(4):486-489.
[19]	JIANG G C, SUN J S, HE Y B, et al. Novel water-based drilling and completion fluid technology to improve wellbore quality during drilling and protect unconventional reservoirs[J]. Engineering, 2021, 18:129-142.
[20]	许洁,许林,李习文,等. 新型储层钻井完井一体化工作液设计及性能评价[J]. 钻井液与完井液,2023,40(2):184-192. doi: 10.12358/j.issn.1001-5620.2023.02.006 XU Jie, XU Lin, LI Xiwen, et al. Design and evaluation of an integrated drilling and completion fluid[J]. Drilling Fluid & Completion Fluid, 2023, 40(2):184-192. doi: 10.12358/j.issn.1001-5620.2023.02.006
[21]	BREIMAN L. Random forest[J]. Machine Learning, 2001, 45(1):5-32. doi: 10.1023/A:1010933404324
[22]	RODGERS J L, NICEWANDER W A. Thirteen ways to look at the correlation coefficient[J]. The American Statistician, 1988, 42(1):59-66.
[23]	KRUSKAL W H. Ordinal measures of association[J]. Journal of the American Statistical Association, 1958, 53(284): 814-861.
[24]	SPEARMAN C. The proof and measurement of association between two things[J]. International Journal of Epidemiology, 2010, 39(5): 1137-1150.