杨丽娜,许胜利,魏莉,史长林,张雨,杨勇
YANG Lina,XU Shengli,WEI Li,SHI Changlin,ZHANG Yu,YANG Yong

• 1:中海油能源发展股份有限公司工程技术分公司
• 2:中海石油(中国)有限公司深圳分公司

摘要(Abstract)：

对不同地震属性裂缝预测体的信息融合是目前碳酸盐岩储层裂缝预测的重难点之一。针对现有信息融合技术中存在的权重系数随机性强、效率低、耗时长、裂缝预测精度不理想等问题，利用机器学习多属性融合方法，基于神经网络系统的单井裂缝解释和多种地震方法的多尺度裂缝预测，得到机器学习融合的训练样本数据集，通过数据编码及结构化处理、标签数据提取及样本集划分和机器学习算法优选等，建立裂缝预测数据驱动模型，对碳酸盐岩储层裂缝智能预测技术进行研究。通过上述方法，得到一个多信息融合的智能裂缝预测强度体，该体能够反映不同尺度裂缝在三维空间的发育强度，反映裂缝各向异性。将技术方法应用至南海流花11-1油田表明，基于机器学习的多属性裂缝融合方法不仅提高工作效率，且有效提高裂缝预测精度，很好地反映裂缝的各向异性，与生产动态特征符合率达90%。研究结果为基于机器学习的高效、高精度多属性裂缝融合预测提供了技术支撑。
Information fusion of fracture prediction bodies with different seismic attributes is one of the major difficulties in current carbonate rock reservoir fracture prediction. In view of the problems existing in present information fusion techniques, such as strong randomness of weight coefficient, low efficiency, long time consumption, and not satisfactory accuracy of fracture prediction, by using machine learning multi-attributes fusion method, training samples data set of machine learning fusion is obtained based on single-well fracture interpretation of neural network system and multi-scales fracture prediction with multiple seismic methods. Through data coding and structural processing, label data extraction, sample set division and machine learning algorithm optimization, data driven model for fracture prediction is established to study intelligent prediction technique for carbonate rock reservoir fractures.Through the above method, a multi information fusion intelligent fracture prediction strength volume is obtained, which can reflect the development intensity of different scale fractures in three-dimensional space and reflect the anisotropy of fractures. The method is applied in Liuhua 11-1 Oilfield in South China Sea,showing that multi-attributes fracture fusion method based on machine learning not only improves work efficiency, but also effectively improves fracture prediction accuracy and well reflects the anisotropy of fractures with coformity of production performance characteristics reaching 90%. The research provides technical support for high-efficiency and high-accuracy multi-attributes fracture fusion prediction based on machine learning.

关键词(KeyWords)： 碳酸盐岩储层;机器学习;多属性融合;裂缝智能预测;单井裂缝解释
carbonate reservoir;machine learning;multi-attributes fusion;intelligent fracture prediction;single-well fracture interpretation

Abstract:

Keywords:

基金项目(Foundation): 中国海洋石油集团有限公司科技攻关项目“双重介质碳酸盐岩油藏调驱/堵控水技术研究与应用”（CNOOC-KJ 135KJXM NFGJ2019-05）;中国海洋石油集团有限公司科技攻关项目“基于深度机器学习的油气储层预测技术”（CNOOC-KJ 135KJXM NFGJ2019-06）

作者(Author): 杨丽娜,许胜利,魏莉,史长林,张雨,杨勇
YANG Lina,XU Shengli,WEI Li,SHI Changlin,ZHANG Yu,YANG Yong

DOI: 10.19597/j.issn.1000-3754.202208009

参考文献(References)：

• [1]刘俊州，韩磊，时磊，等.致密砂岩储层多尺度裂缝地震预测技术：以川西XC地区为例[J].石油与天然气地质，2021,42(3):747-754.LIU Junzhou, HAN Lei, SHI Lei, et al. Seismic prediction of tight sandstone reservoir fractures in XC area, western Sichuan Basin[J]. Oil&Gas Geology,2021,42(3):747-754.
• [2]朱志强.利用数字岩心技术研究变质岩潜山裂缝油藏剩余油特征[J].特种油气藏，2019,26(3):148-152.ZHU Zhiqiang. Study on remaining oil characteristics of fractured reservoir in metamorphic rock buried hill by digital core technology[J].Special Oil&Gas Reservoirs,2019,26(3):148-152.
• [3]房娜，姜光宏，李云鹏，等.潜山裂缝油藏开发特征及挖潜方向[J].特种油气藏，2017,24(3):90-94.FANG Na, JIANG Guanghong, LI Yunpeng, et al. Development characteristics and potential tapping direction of buried hill fractured reservoir[J]. Special Oil&Gas Reservoirs, 2017,24(3):90-94.
• [4]王晓阳，张晓斌，赵晓红，等.四川盆地复杂地表地震采集关键技术及其应用效果[J].天然气工业，2021,41(7):15-23.WANG Xiaoyang, ZHANG Xiaobin, ZHAO Xiaohong, et al.Key technologies for complex surface seismic acquisition in the Sichuan Basin and their application effect[J]. Natural Gas Industry,2021,41(7):15-23.
• [5]黄浩勇，苟其勇，刘胜军，等.页岩气叠后裂缝综合预测技术：以长宁国家级页岩气示范区为例[J].断块油气田，2022,29(2):218-223.HUANG Haoyong, GOU Qiyong, LIU Shengjun, et al. Poststack fracture comprehensive prediction technology for shale gas:Taking Changning national shale gas demonstration area as an example[J]. Fault-Block Oil&Gas Field, 2022, 29(2):218-223.
• [6]刘学义，程玖兵，王腾飞，等.低信噪比海底多分量地震数据波场分离方法[J].地球物理学报，2021,64(2):684-699.LIU Xueyi,CHENG Jiubing,WANG Tengfei,et al. Wave field separation method for low signal-to-noise ratio seabed multi-component seismic data[J]. Chinese Journal of Geophysics,2021,64(2):684-699.
• [7]范存辉，顿雅杭，张玮，等.准噶尔盆地中拐凸起火山岩储集层裂缝综合评价[J].新疆石油地质，2017, 38(6):693-700.FAN Cunhui, DUN Yahang, ZHANG Wei, et al. Comprehensive evaluation of fractures in volcanic reservoirs of Zhongguai Swell,Junggar Basin[J]. Xinjiang Petroleum Geology,2017,38(6):693-700.
• [8]束宁凯，汪新文，宋亮，等.基于地震影像学的砂砾岩体多属性融合裂缝预测：以济阳坳陷车西地区北部陡坡带沙三段为例[J].油气地质与采收率，2016,23(2):57-61.SHU Ningkai, WANG Xinwen, SONG Liang, et al. Fracture prediction of sand-gravel body by multi-attribute fusion based on seismic imaging theory:A case study of the third member of Shahejie Formation in northern zone of Chexi area in Jiyang depression[J]. Petroleum Geology and Recovery Efficiency, 2016, 23(2):57-61.
• [9]于晓东，桂志先，汪勇，等.叠前AVAZ裂缝预测技术在车排子凸起的应用[J].石油地球物理勘探，2019,54(3):624-633.YU Xiaodong,GUI Zhixian,WANG Yong,et al. Application of prestack AVAZ fracture prediction technology in Chepaizi uplift[J]. Oil Geophysical Prospecting,2019,54(3):624-633.
• [10]卢溜.地震多属性分析在致密储层裂缝识别中的应用[D].青岛：中国石油大学（华东），2017.LU Liu.Application of seismic multi-attribute analysis in fracture identification of tight reservoir[D]. Qingdao:China University of Petroleum(East China),2017.
• [11]郭子南.基岩潜山内幕裂缝测井响应特征研究[J].中国石油和化工标准与质量，2021,41(1):117-119.GUO Zinan. Study on logging response characteristics of internal fractures in bedrock buried hill[J]. China Petroleum and Chemical Standards and Quality,2021,41(1):117-119.
• [12]杨华，张本健，陈明江，等.裂缝性碳酸盐岩储层气水识别的测井解释新途径：以四川盆地九龙山地区中二叠统为例[J].天然气工业，2021,41(7):56-62.YANG Hua,ZHANG Benjian,CHEN Mingjiang,et al. A new log interpretation method for identifying water and gas layers in fractured carbonate reservoirs:A case study of Middle Permian in Jiulongshan area of northwestern Sichuan Basin[J]. Natural Gas Industry,2021,41(7):56-62.
• [13]杨曦.致密储层裂缝地质成因、特征及测井专属响应与表征[D].北京：中国地质大学（北京），2021.YANG Xi. Geological origin,characteristics and logging specific response and characterization of fractures in tight reservoirs[D].Beijing:China University of Geosciences(Beijing),2021.
• [14]李宏波，罗平亚，白杨，等.机器学习算法概述及其在钻井工程中的应用[J].新疆石油天然气，2022,18(1):1-13.LI Hongbo, LUO Pingya, BAI Yang, et al. Overview of machine learning algorithm and its application in drilling engineering[J]. Xinjiang Oil&Gas,2022,18(1):1-13.
• [15]张仕斌，黄曦，昌燕，等.大数据环境下量子机器学习的研究进展及发展趋势[J].电子科技大学学报，2021,50(6):802-819.ZHANG Shibin, HUANG Xi, CHANG Yan, et al. Research progress and development trend of quantum machine learning in big data environment[J]. Journal of University of Electronic Science and Technology of China,2021,50(6):802-819.
• [16]赵明梅，金洋洋，王语嘉，等.随机森林算法在选择决策中的应用研究[J].计算机与网络，2021,47(22):56-59.ZHAO Mingmei,JIN Yangyang,WANG Yujia,et al. Research on application of stochastic forest algorithm in selection decision[J]. Computer and Network,2021,47(22):56-59.