王庆宇,朱伟,李浩,孟丽丽,宋玉梅,王春蕊
WANG Qingyu,ZHU Wei,LI Hao,MENG Lili,SONG Yumei,WANG Chunrui

• 1:中国石油大庆油田有限责任公司勘探开发研究院
• 2:中国石油大庆油田有限责任公司
• 3:中国石油大庆油田有限责任公司第五采油厂
• 4:中国石油大庆油田有限责任公司第二采油厂

摘要(Abstract)：

针对传统地层对比存在主观性强、效率低等问题，为了避免差错、提高工作效率，实现新井地层划分及老井成果质控全过程智能化，基于传统知识构建的知识库，以空洞卷积语义分割模型V3+(DeepLabV3plus)智能算法为核心，形成“知识+数据”双驱动的智能地层对比方法，实现标志层控制下的沉积单元自动对比。结果表明：通过智能算法改进、扩充样本集的规模与多样性、分段优化模型搭建，经过多轮迭代训练与模型质量评估，实现了传统对比与智能算法的有效融合，地层对比预测模型泛化能力显著增强；在南2―3区实例应用中，构建模型训练及验证精度达到90%以上，在人工质控的基础上，智能地层对比准确率再提升1百分点，工作效率可提升10～20倍。研究成果在提升地层对比精度、提高工作效率方面具有重要的应用价值。
In view of the problems of strong subjectivity and low efficiency in traditional stratigraphic correlation, the intelligentization of the whole process of stratigraphic division for new wells and quality control for old wells achievements is realized to avoid errors and improve work efficiency. Based on the knowledge base constructed by traditional knowledge and taking intelligent algorithm of atrous convolution semantic segmentation(DeepLabV3plus) as the core, an intelligent stratigraphic correlation method driven by “knowledge + data” is developed, realizing the automatic correlation for sedimentary units controlled by marker horizons. The results show that, through improvements in the intelligent algorithm, expansion of sample-set scale and diversity and segmented optimization of model construction, multiple rounds of iterative training and model quality evaluation effectively integrate the traditional comparison and intelligent algorithm, with generalization capability of the stratigraphic correlation prediction model significantly improved. Case application in South 2-3 area indicates that, the training and verification accuracy of the constructed model is >90 %. On the basis of manual quality control, the accuracy of intelligent stratigraphic correlation increases by additional 1 percentage point, while the work efficiency increases by 10-20 times. The research provides important application value in improving stratigraphic correlation accuracy and work efficiency.

关键词(KeyWords)： 智能地层对比;DeepLabV3plus算法;标志层;分段模型;模型评估
intelligent stratigraphic correlation;DeepLabV3plus algorithm;marker horizon;segmented model;model evaluation

Abstract:

Keywords:

基金项目(Foundation): 中国石油天然气股份有限公司攻关性应用型科技专项“中高渗油田特高含水期大幅度提高采收率技术研究”(2023ZZ22)

作者(Author): 王庆宇,朱伟,李浩,孟丽丽,宋玉梅,王春蕊
WANG Qingyu,ZHU Wei,LI Hao,MENG Lili,SONG Yumei,WANG Chunrui

DOI: 10.19597/J.ISSN.1000-3754.202504054

参考文献(References)：

• [1]刘建伟，丁熙浩，罗雄麟.多模态深度学习综述[J].计算机应用研究，2020,37(6):1601-1614.LIU Jianwei,DING Xihao,LUO Xionglin.Survey of multimodal deep learning[J]. Application Research of Computers,2020,37(6):1601-1614.
• [2]邬德刚，吴胜和，刘磊，等.基于模式约束的油层单元智能自动对比方法：以盆地史南油田史深100区块加积式地层对比为例[J].石油勘探与开发，2024,51(1):161-172.WU Degang,WU Shenghe,LIU Lei,et al. An intelligent automatic correlation method of oil-bearing strata based on pattern constraints:An example of accretionary stratigraphy of Shishen 100block in Shinan Oilfield of Bohai Bay Basin, East China[J].Petroleum Exploration and Development, 2024, 51(1):161-172.
• [3]肖波，韩学辉，周开金，等.测井曲线自动分层方法回顾与展望[J].地球物理学进展，2010,25(5):1802-1810.XIAO Bo,HAN Xuehui,ZHOU Kaijin,et al. A review and outlook of automatic zonation methods of well log[J]. Progress in Geophysics,2010,25(5):1802-1810.
• [4]邵广辉，高衍武，张苏利，等.变窗重建算法和自适应SegNet网络在地层层序划分中的应用[J].长江大学学报（自然科学版），2024,21(4):19-31.SHAO Guanghui,GAO Yanwu,ZHANG Suli, et al. Variable window reconstruction algorithm and adaptive SegNet network in the application of stratigraphic sequence classification[J]. Journal of Yangtze University(Natural Science Edition), 2024,21(4):19-31.
• [5]崔国宏，郭明宇，苑仁国，等.地层自动对比预测方法研究与应用[J].科技创新与应用，2023,13(6):155-165.CUI Guohong, GUO Mingyu, YUAN Renguo, et al. Research and application of automatic stratigraphic correlation prediction method[J]. Technological Innovation and Application, 2023,13(6):155-165.
• [6]马陇飞，萧汉敏，陶敬伟，等.基于深度学习岩性分类的研究与应用[J].科学技术与工程，2022,22(7):2609-2617.MA Longfei,XIAO Hanmin,TAO Jingwei,et al. Research and application of lithology classification based on deep learning[J]. Science Technology and Engineering, 2022, 22(7):2609-2617.
• [7]徐少华，张宇航，宋美玲，等.基于特征识别与PSO结合的地层对比算法[J].计算机技术与发展，2015,25(5):37-40.XU Shaohua,ZHANG Yuhang,SONG Meiling,et al. A stratigraphic correlation algorithm based on characteristics identification and PSO[J]. Computer Technology and Development,2015,25(5):37-40.
• [8] YANG Y,WANG J Y,LI Z,et al. Automated stratigraphic correlation of well logs using attention based dense network[J].Artificial Intelligence in Geosciences,2023,4(1):128-136.
• [9] CHEN B Y, ZENG X J, ZHANG B Y, et al. Interwell stratigraphic correlation detection based on knowledge-enhanced fewshot learning[J]. Geoenergy Science and Engineering, 2023,220:111187.
• [10] QIU Q J,DUAN Y X,MA K,et al. Information extraction and knowledge linkage of geological profiles and related contextual texts from mineral exploration reports for geological knowledge graphs construction[J]. Ore Geology Reviews, 2023, 163:105739.
• [11] MUHAMMAD T N,SHAZIA N,MUHAMMAD A S. Simulating the stratigraphy of meandering channels and point bars of Cretaceous system using spectral decomposition tool,Southwest Pakistan:Implications for petroleum exploration[J]. Journal of Petroleum Science and Engineering,2022,212:110201.
• [12] GONG W P,ZHAO C,JUANG C H,et al. Stratigraphic uncertainty modelling with random field approach[J]. Computers and Geotechnics,2020,125:103681.
• [13]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，2005,24(1):5-9,12.ZHAO Hanqing. High-resolution sequential stratigraphy correlation and Chinese subzone correlation[J]. Petroleum Geology&Oilfield Development in Daqing,2005,24(1):5-9,12.
• [14]刘波.基准面旋回与沉积旋回的对比方法探讨[J].沉积学报，2002,20(1):112-117.LIU Bo. Discussion on the correlation methods of base-level cycle and sedimentary cycle sequence[J]. Acta Sedimentologica Sinica,2002,20(1):112-117.
• [15]何宇航，王庆宇，朱伟，等.基于密井网水下分流河道单砂体原型地质模型[J].大庆石油地质与开发，2018,37(3):43-48.HE Yuhang,WANG Qingyu,ZHU Wei,et al. Prototype geological model for the individual sandbody of the underwater distributary channel based on the dense well pattern[J]. Petroleum Geology&Oilfield Development in Daqing, 2018, 37(3):43-48.
• [16]付志国，郑荣才，赵翰卿，等.体积分配原理在河流-三角洲相储层分析中的应用：以大庆长垣萨葡油层为例[J].成都理工大学学报（自然科学版），2006,33(5):504-508.FU Zhiguo,ZHENG Rongcai,ZHAO Hanqing,et al.Application of capacity distribution principle in river-delta reservoir analyses:Taking the Daqing Changyuan Sapu oil layer as example[J]. Journal of Chengdu University of Technology(Science&Technology Edition),2006,33(5):504-508.
• [17]阮壮，朱筱敏，何宇航，等.大庆长垣北部葡萄花上部油层高分辨率层序地层划分[J].沉积学报，2012,30(2):301-309.RUAN Zhuang,ZHU Xiaomin,HE Yuhang,et al. High-resolution sequence stratigraphic division of upper Putaohua reservoir,northern Daqing Placanticline[J]. Acta Sedimentologica Sinica,2012,30(2):301-309.
• [18]于德水，何宇航，邢宝荣，等.大庆长垣高台子油层沉积演化分布及沉积模式[J].沉积学报，2024,42(1):238-250.YU Deshui,HE Yuhang,XING Baorong,et al. Study of evolution,distribution and sedimentary model of Gaotaizi reservoir in Daqing Placanticline[J]. Acta Sedimentologica Sinica, 2024,42(1):238-250.
• [19]杨云，余逸凡，毛平.复杂地层精细地层对比方法：以尕斯N1-N21油藏为例[J].石油地质与工程，2010, 24(4):22-26.YANG Yun,YU Yifan,MAO Ping.Fine stratigraphic correlation and its application in complicated formation:A case study of Gasi N1-N21 reservoir[J]. Petroleum Geology and Engineering,2010,24(4):22-26.
• [20]黄勇，杨会东，李艳春，等.油田密井网条件下井震匹配构造表征技术及应用[J].中国石油大学学报（自然科学版），2023,47(4):60-68.HUANG Yong,YANG Huidong,LI Yanchun,et al. Structure characterization technology and application of well seismic matching under dense well pattern in oilfield[J]. Journal of China University of Petroleum(Edition of Natural Science), 2023,47(4):60-68.
• [21]赵春晨，刘豪.基准面旋回控制下的浅水三角洲砂体分散体系特征[J].新疆石油地质，2023,44(6):657-666.ZHAO Chunchen,LIU Hao. Characteristics of sand body dispersion system in shallow-water delta controlled by base-level cycle[J]. Xinjiang Petroleum Geology,2023,44(6):657-666.