Research Progress and Development Prospect of Intelligent Surface Logging Technology
-
摘要:
录井具有样品条件及制样工序复杂、采集项目多而离散、人工经验依赖性强且人均产值低等特点，亟需加强智能化转型，但相比于其他石油工程技术，智能录井技术进展缓慢，且局限于应用层面。为此，从智能钻井的进展与成效入手，分析了国内外智能钻井在硬件系统、控制系统、应用系统方面的进展与差距；然后，从地质录井、工程录井、智慧平台3个方面分析了智能录井的主要技术进展，包括“数据+”驱动和视觉驱动的岩性识别、流体识别、井下与地面风险识别及预警等。通过对比智能钻井与智能录井的现状，提出智能录井应强化井下智能录井、智能录井机器人等硬件系统及多场数字孪生、多元采集智能控制、多模态录井大模型、智能解释评价等软件系统的研发。同时强调，既要高度重视，又要理性看待智能录井的发展，要在回顾评价、横向对比的基础上，做好战略定位与研发流程优化，实现进度追赶与作用发挥。这些分析与观点，对推动智能录井实现良性、快速发展具有指导意义。
Abstract:Surface logging has the characteristics of complex sample conditions and sample preparation process, numerous and discrete collection items, strong dependence on hands-on experience, and low per capita output. It is urgent to strengthen intelligentialization transformation. However, compared with other petroleum engineering technologies, intelligent surface logging is making slow advancement and facing application limits. Therefore, the progress and gap of hardware systems, control systems, and application systems of intelligent drilling in China and abroad were analyzed from the progress and achievement of intelligent drilling. Then, the main technical progress of intelligent surface logging was analyzed in terms of geological surface logging, engineering surface logging, and intelligent platform, covering “data +” driven and visually driven lithology identification, fluid identification, and downhole and surface risk identification and early alarming. Based on the comparison between intelligent drilling and intelligent surface logging, it was suggested that the research and development of hardware systems such as downhole intelligent surface logging, and intelligent surface logging robots, as well as software systems such as multi-field digital twins, multi-acquisition intelligent control, multimode large model of surface logging, and intelligent interpretation and evaluation should be strengthened. At the same time, it was emphasized that we should attach great importance to and rationally look at the development of intelligent surface logging and determine strategic positioning and process optimization on the basis of retrospective evaluation and horizontal comparison, so as to catch up with the progress and engagement. These analyses and viewpoints have guiding significance in promoting the benign and rapid development of intelligent surface logging.
-
Keywords:
- Intelligent surface logging /
- intelligent drilling /
- robots /
- digital twin /
- large model /
- intelligent control
-
水平井桥塞分段压裂已成为非常规油气高效开发的核心技术之一[1]。桥塞用于封隔已压裂井段裂缝，通常桥塞中心管中空，需井口投球、落入桥塞球座，起到密封桥塞中心管的作用。当桥塞球座处于完全密封状态，则为有效坐封[2]。一旦球座坐封失效，易发生压裂砂堵、重复压裂已改造井段等问题，无法正常后续作业。目前主要依靠人为观察井口压力变化特征判断坐封的有效性，然而对于非典型压力特征，难以快速准确识别，这已成为制约水平井桥塞分段压裂技术发展的瓶颈之一。
近年来，以大数据、机器学习、超强算力为基础的新一代人工智能技术蓬勃发展[3–4]，基于海量压裂历史数据，通过人工智能算法从大数据中学习数据变化特征[5–7]，形成了压裂工况智能诊断方法，达到实时诊断压裂工况的目的。前人已开展基于大数据分析的压裂起止时刻、暂堵、球座坐封等工况诊断研究。A. Ramirez等人[8]采用分类算法结合泵压曲线和专家经验，实现压裂作业起始与终止时刻的识别；M. M. Awad等人[9]利用小波变换方法将施工泵压蕴含的能量信息与裂缝扩展物理过程相关联，实现了单位时间内裂缝扩展事件数的定量表征；袁彬等人[10]结合长短期记忆神经网络、反向传播神经网络等多种模型，实现了泵球、前置酸降压、暂堵压裂、砂堵等事件的智能识别；盛茂等人[11]利用聚类算法、特征参数阈值法分析压裂施工数据，建立了暂堵有效性评价模型。Shen Yuchang等人[12]利用包含地面泵压以及排量的施工曲线图，基于识别图像的U-Net架构深度学习算法，建立了桥塞球座坐封起止时刻的识别模型，识别准确率达95%。该研究的判别特征是单一的排量下降，而复杂地层压裂作业过程中往往存在大量的排量下降现象，但这些并不都是由桥塞球座坐封造成的，因此仅以排量下降作为识别坐封工况的特征具有一定局限性。
为此，笔者融合专家经验定性判识和坐封数据特征挖掘定量标注，滑动窗口数据切片形成5 792组样本数据，优选井口压力–排量二维输入的长短期记忆神经网络，建立了压裂球座坐封有效性智能诊断模型；并采用欠采样平衡数据集方式提升模型判识精度，实现了每秒输出诊断结果，为桥塞球座坐封有效性实时自动诊断提供了方法。
1. 球座坐封特征参数提取
1.1 压裂球座坐封工况数据分割与标签标注
水平井桥塞分段压裂时，压裂投球坐封阶段，排量先降至0.5～1.0 m3/min，维持压裂球以较低速度坐入桥塞球座；当球座被完全密封，井筒内流体憋压，此时井口压力显著升高[13]；压力达到地层破裂压力使地层破裂后，井口压力骤降；随后逐级提高排量至压裂设计值，井口压力缓慢上升，整个投球坐封阶段井口压力呈现陡升—陡降—平缓上升的显著特征（见图1）。根据该特征，采用滑动窗口方式，对球座坐封工况进行数据分割：在排量开始降低且累计排量小于井筒体积的时段寻找第一次压力突升的时刻，并标记为坐封开始时刻；在滑动时间窗口寻找压力降落结束时刻，向后继续寻找至排量开始增加的时刻，并标记为坐封结束时刻。
工况分割后形成198段坐封数据，结合专家经验对井口压力和排量的变化特征进行判识，对每段坐封数据打标签，分为有效坐封和无效坐封2类标签，分别将其对应的时刻标记为数字1和0，最终得到有效坐封168段，无效坐封30段。
1.2 有效坐封数据特征分析
整合并提取168段有效坐封数据共有的特征点，分别是压力升高前的最小值、压力升高后达到的峰值、压力下降后的最小值及排量开始增加时刻的压力回升值，形成1—2，2—3，3—4明显的三阶段特征（见图2（a））；统计3个阶段对应的持续时间及井口压力变化值，得到各阶段的斜率分布和小提琴图（见图2（b）和（c））。
![]()
图 2 井口压力三阶段变化特征及斜率分布情况Figure 2. Characteristics of wellhead pressure change and slope distribution in three stages从图2可知，阶段一压力上升变化斜率区间[0.14，5.62]；阶段二压力下降变化斜率区间 [−0.01，−7.39]，分布范围较大；阶段三压力上升变化斜率区间[0.02，1.36]，与前2个阶段相比，分布区间更为集中。尽管有效坐封数据均呈显著的井口压力陡升—陡降—平缓上升的三阶段特征，但是数据样本分布范围较大、且不一致，无法形成明确的诊断规则实现准确诊断。
1.3 无效坐封数据特征分析
统计分析30段无效坐封数据，发现存在2类形态的无效坐封曲线（见图3）。其中，A类无效坐封虽然存在阶段一对应的压力升高过程，但阶段二的压力会降至低于压力升高前的起始值；B类无效坐封的井口压力在降低后直至提排量前始终未呈现回升趋势。
![]()
图 3 坐封失效时对应的2类井口压力曲线Figure 3. Two types of wellhead pressure curves corresponding to setting failure2. 基于长短期记忆神经网络模型建立
2.1 神经网络结构设计
压裂泵注曲线具有显著的长时序性和数据关联性，即压裂全过程具有较长的时间跨度，且每个时间步长之间的压力存在依赖关系，为此优选长短期记忆神经网络[14–18]，其特有的记忆门控单元可捕捉序列数据中的长期依赖关系[19–20]。神经网络设计采用二分类问题的设计思路，其中隐含层初始设置为256层，输入维度分别设置为一维和二维，输出设置为代表坐封有效与失效的1或0标签，对每一种输入特征值进行批标准化处理；选择Softmax激活函数对每一维度相同位置的数值进行Softmax运算，每次模型调用时对待训练参数矩阵和待训练偏置项进行初始化处理；选用Adam优化器处理二分类问题，初始学习率设置为0.01；使用交叉熵损失函数表征模型输出的有效坐封标签与实际有效坐封标签的偏差，来衡量该网络在此数据集上对坐封有效性识别效果的好坏。每次输入神经网络的训练集样本数初始设置为128个，迭代200次，每次迭代输出一次损失函数，保留最后一次训练参数，并计算准确率。神经网络结构如图4所示。
![]()
图 4 二维输入的长短期记忆神经网络结构Figure 4. Structure of LSTM neural network with two-dimensional input图4中，t表示时间，p和Q分别表示井口压力与排量的时序值，X和Y表示LSTM网络的输入和输出，σ和tanh分别表示sigmoid激活函数和tanh激活函数，参数C、h、f、i和o分别代表LSTM网络的细胞状态、隐藏状态、遗忘门、输入门和输出门。
2.2 标签数据切片处理
全连接层的神经元网络用分割出的坐封段作为训练数据集时，训练集太过冗长，会大大增加迭代时间，影响训练效果[21]，因此采用滑动窗口切片送入模型的方式来缩短训练的数据长度[22]。
统计每段数据中有效坐封的3个阶段特征的时长，得出坐封所需的时间最长为252 s。为保证特征被全部包含，将时间跨度增加，初步设定切片窗口为300 s。考虑到时间序列前后数据的相关性，设置移动步长为50 s，即窗口每次向后移动50 s，以保证对同样的一段坐封特征，其前后段的数据都能作为有效坐封的样本输入，同时也增加了样本数量。窗口从左到右，依次对每个时刻的标签进行扫描，当该窗口第一秒和最后一秒的标签均为0，且中间有且仅有一段连续为1的标签时，视为此窗口包含了一个完整的坐封段，并将整个窗口标记为标签1，作为一个有效坐封样本；当窗口内数据标签均为0，或者由1开始与结束，即无坐封段或坐封段不完整时，整个窗口标记为标签0，作为一个无效坐封样本。滑动窗口切片标注如图5所示。
198段坐封工况数据按照时间窗口300 s、移动步长50 s切片后形成5 792个样本，其中有效坐封383个，占比仅6.61%。当二分类模型中标签为1的数据量极少时，神经网络被重复传入大量的无效坐封样本，从而无法学习到有效坐封的特征。为此，采用欠采样平衡数据集方式[23]，从切片后的样本中等比例地提取标签为1和0的样本，总计766个，再以8∶2的比例划分为训练集和测试集，最终形成610个样本的训练集，156个样本的测试集。
3. 模型训练与结果分析
为考察压力和排量变化对坐封有效性判识效果的影响，分别建立井口压力一维输入和井口压力–排量二维输入的长短期记忆神经网络模型进行对比训练。首先，调整数据切片时间窗口为300，400和500 s，随着时间窗口增长，准确率由88%降至70%，表明过长的时间窗口导致样本包含更多冗余的数据信息，从而输入的干扰特征增多，因此时间窗口选择300 s；然后，调整批量大小为64，128和256，学习率分别为0.001，0.01和0.1，进行组合训练，训练结果如图6（a）所示。训练结果表明，批量大小为256时，模型准确率整体偏低，仅为50%～70%；批量大小为64、学习率为0.001时，井口压力一维输入模型的准确率最高为91.7%；批量大小为128、学习率为0.01时，井口压力–排量二维输入模型的准确率最高为96.8%，相比井口压力一维输入模型提高5.1百分点。2种模型准确率最高时对应的损失函数变化曲线如图6（b）所示。由图6（b）可以看出，迭代至第25次时，井口压力一维输入模型的损失函数降至0.30，而井口压力–排量二维输入模型的损失函数降至0.15，收敛速度更快，最终趋近于0.10。
验证集选用长庆油田合水区块51段未参与训练的压裂数据。将井口压力和排量数据以时间窗口300 s、移动步长1 s滑动输入模型，模型调用训练准确率最高的权重参数进行判识，实时输出坐封工况判识标签。若井口压力从排量降至送球排量时开始运行到累计液量达到一个井筒体积时仍未呈现三阶段特征，则判识为无效坐封。对比专家经验标签，井口压力一维输入的模型准确率为73.7%，井口压力–排量二维输入的模型准确率为84.3%。将模型识别出的43段有效坐封段绘制成瀑布图（见图7（a）），可以看出，虽然井口压力数据跨度分布较大，但该模型均能正确判识，验证了模型的有效性；将实际有效坐封、但模型误判为无效坐封的8段数据绘制成瀑布图（见图7（b）），发现此类曲线在压力突升至峰值后有一段时间的缓慢爬升，未能被该模型识别，其原因是此类情况下输入样本不足，长短期记忆神经网络未能学习到该类曲线的特征。
4. 结 论
1）针对压裂桥塞球座坐封有效性难以形成有效规则、快速准确判识的问题，提出了人工智能技术辅助诊断方法。融合专家经验定性判识和坐封数据特征挖掘定量标注，建立了基于长短记忆神经网络的压裂球座坐封有效性智能诊断模型，并采用欠采样平衡数据集方式提高模型的预测精度。
2）压裂桥塞球座有效坐封时，井口压力呈现陡升—陡降—平缓上升特征，但各阶段变化值分布离散，持续时间跨度大；无效坐封时，井口压力呈2种形态，一种是陡降幅度超过陡升幅度，一种是缺少平缓上升阶段特征。
3）利用未参与训练的51个样本验证模型，井口压力–排量二维输入模型成功识别出43个有效坐封段，准确率达84.3%。
-
表 1 录井与钻井智慧平台架构的对比
Table 1 Comparison of intelligent platform architecture between drilling and logging
层次 智能录井[64] 智能钻完井[1] 钻井[28] 用户层 移动端、Pad端、PC端的实时监控、技术支持、协同研究、远程决策 钻井工程师、完井工程师、地质工程师、管理人员、管理员 网络层 局域网、广域网 应用层 数据管理平台（分类存储与查询等）、数据挖掘平台（统计分析方法）、成果输出平台（岩性自动识别、物性评价、油气层解释、工程智能预警等） 机械钻速智能预测与参数优化、井眼轨迹智能优化与闭环调控、钻井风险智能预警与动态调控、固井质量智能评价与优化控制、压裂方案智能设计和优化调控、完井方案智能设计与生产优化及钻完井多过程动态耦合与多目标协同优化 钻头选型、井壁稳定、钻速预测、卡钻预警、钻井参数优化等 装备层 智能钻头、井下测量短节、智能导向工具、智能钻杆、智能滑套、智能钻机 算法层/数据操作层 机理数据融化、数据增强、小样本学习、迁移学习、强化学习、卷积神经网络、小波分析、在线学习、图算法、遗传算法 数据清洗、资源调度、计算工程 数据层 各类录井仪器、传感器采集的数据、图像、音频/视频等信息 物探数据、综合录井数据、测井数据、岩心数据、地质资料数据、随钻数据、文档资料、其他数据 井信息、录井、测井、地质、地震实时级历时数据 
下载: 导出CSV
表 2 录井信息处理与解释评价技术体系
Table 2 Technical system of logging information processing, interpretation, and evaluation
录井信息处理 解释 评价及应用 深度校正与数据源深度匹配
影响因素校正或消除
散失量恢复或原位重构
解谱解耦与信息挖掘
曲线、图谱、影像特征提取
多元、多维、多尺度数据融合
平滑、抽稀、插值等处理
标准化、归一化处理岩性、岩相识别
古生物鉴定及沉积环境识别
成分及结构、构造识别
地质层位及地质小层识别
流体类型及赋存状态识别
物源、油（气）源识别
油（气）成因识别
含水性及水型识别
有效储层识别
油气水层解释
VOCs类型识别
钻井工况与安全风险识别
高压层及其成因识别
溢流预警
井漏及其原因识别物性及孔隙结构评价
烃源岩特性评价
脆性、岩石力学、可压性评价
润湿性、水淹层评价
含油气丰度及油气性质评价
含水性或含水率评价
甜点评价或产能预测
单井评价或选区评价
地层压力随钻评价
可钻性、井壁稳定性评价
井筒封闭性或盖层评价
热储及锂、钾、铀丰度评价
钻井地质设计
水平井综合地质导向
压裂选层
下载: 导出CSV
-
[1] 李根生,宋先知,祝兆鹏,等. 智能钻完井技术研究进展与前景展望[J]. 石油钻探技术,2023,51(4):35–47. doi: 10.11911/syztjs.2023040 LI Gensheng, SONG Xianzhi, ZHU Zhaopeng, et al. Research progress and the prospect of intelligent drilling and completion technologies[J]. Petroleum Drilling Techniques, 2023, 51(4): 35–47. doi: 10.11911/syztjs.2023040
[2] 王敏生,光新军. 智能钻井技术现状与发展方向[J]. 石油学报,2020,41(4):505–512. doi: 10.7623/syxb202004013 WANG Minsheng, GUANG Xinjun. Status and development trends of intelligent drilling technology[J]. Acta Petrolei Sinica, 2020, 41(4): 505–512. doi: 10.7623/syxb202004013
[3] 李宗田,肖勇,李宁,等. 低油价下的页岩油气开发工程技术新进展[J]. 断块油气田,2021,28(5):577–585. LI Zongtian, XIAO Yong, LI Ning, et al. New progress in shale oil and gas development engineering technology under low oil prices[J]. Fault-Block Oil & Gas Field, 2021,28(5): 577–585.
[4] MCCARTHY J, MINSKY M L, ROCHESTER N, et al. A proposal for the Dartmouth summer research project on artificial intelligence: August 31, 1955[J]. AI Magazine, 2006, 27(4): 12–14.
[5] GAINITDINOV B, MESHALKIN Y, ORLOV D, et al. Predicting mineralogical composition in unconventional formations using machine learning and well logging data[R]. IPTC 23487, 2024.
[6] YANG Tao, ARIEF I H, NIEMANN M, et al. A machine learning approach to predict gas oil ratio based on advanced mud gas data[R]. SPE 195459, 2019.
[7] 陈凯枫,杨学文,宋先知,等. 基于工程录井数据的井漏智能诊断方法[J]. 石油机械,2022,50(11):16–22. CHEN Kaifeng, YANG Xuewen, SONG Xianzhi, et al. An intelligent diagnosis method for lost circulation based on engineering logging data[J]. China Petroleum Machinery, 2022, 50(11): 16–22.
[8] 刘枫. 顺北地区地层四压力智能预测软件研发与应用[D]. 北京:中国石油大学(北京),2023. LIU Feng. Research and application of intelligent prediction software for formation four pressure in Shunbei Area[D]. Beijing: China University of Petroleum(Beijing), 2023.
[9] 匡立春,刘合,任义丽,等. 人工智能在石油勘探开发领域的应用现状与发展趋势[J]. 石油勘探与开发,2021,48(1):1–11. doi: 10.11698/PED.2021.01.01 KUANG Lichun, LIU He, REN Yili, et al. Application and development trend of artificial intelligence in petroleum exploration and development[J]. Petroleum Exploration and Development, 2021, 48(1): 1–11. doi: 10.11698/PED.2021.01.01
[10] 王志战. 一体化、智能化时代的录井技术发展方向探讨[J]. 录井工程,2020,31(1):1–6. doi: 10.3969/j.issn.1672-9803.2020.01.001 WANG Zhizhan. Discussion on the development direction of mud logging technology in the era of integration and intellectualization[J]. Mud Logging Engineering, 2020, 31(1): 1–6. doi: 10.3969/j.issn.1672-9803.2020.01.001
[11] 王志战. 中国石化录井技术新进展与发展方向思考[J]. 石油钻探技术,2023,51(4):124–133. doi: 10.11911/syztjs.2023027 WANG Zhizhan. Thoughts for new progress and development directions of Sinopec's surface logging technology[J]. Petroleum Drilling Techniques, 2023, 51(4): 124–133. doi: 10.11911/syztjs.2023027
[12] 闫铁,许瑞,刘维凯,等. 中国智能化钻井技术研究发展[J]. 东北石油大学学报,2020,44(4):15–21. doi: 10.3969/j.issn.2095-4107.2020.04.003 YAN Tie, XU Rui, LIU Weikai, et al. Research and development of intelligent drilling technology in China[J]. Journal of Northeast Petroleum University, 2020, 44(4): 15–21. doi: 10.3969/j.issn.2095-4107.2020.04.003
[13] 张鑫鑫,梁博文,张晓龙,等. 智能钻井装备与技术研究进展[J]. 煤田地质与勘探,2023,51(9):20–30. doi: 10.12363/issn.1001-1986.23.06.0324 ZHANG Xinxin, LIANG Bowen, ZHANG Xiaolong, et al. Research progress of intelligent drilling equipment and technology[J]. Coal Geology & Exploration, 2023, 51(9): 20–30. doi: 10.12363/issn.1001-1986.23.06.0324
[14] HU Qin, LIU Qingyou. Intelligent drilling: a prospective technology of tomorrow[R]. SPE 103781, 2006.
[15] RASSENFOSS S. Drilling automation: a robot takes over the drilling floor[J]. Journal of Petroleum Technology, 2021, 73(12): 18–22. doi: 10.2118/1221-0018-JPT
[16] LAWRENCE L, REDMOND B, RUSSELL R B, et al. Intelligent wired drill-pipe system provides significant improvements in drilling performance on offshore Australia development[R]. OTC 20067, 2009.
[17] JELLISON M J, PRIDECO G, HALL D R. Intelligent drill pipe creates the drilling network[R]. SPE 80454, 2003.
[18] TURNER D R, HEAD P F, YURATICH M A, et al. The all electric BHA: recent developments toward an intelligent coiled-tubing drilling system[R]. SPE 54469, 1999.
[19] VALVERDE E. Intelligent near-bit reamer affords same-trip drilling, hole enlargement and rathole reduction for optimal deepwater well construction[R]. OTC 28402, 2018.
[20] TILLEY J, NAIR V N, HAMOUDI L. Case study: intelligent RSS improving drilling performance on three mile laterals in the Appalachian Basin[R]. SPE 201723, 2020.
[21] AL ARFI S, ALSOWAIDI F, RUIZ F, et al. New intelligent push-the-bit rotary steerable system helped reducing well time and maximized directional drilling performance, Abu Dhabi, UAE[R]. SPE 207537, 2021.
[22] CORSER G P, HARMSE J E, CORSER B A, et al. Field test results for a real-time intelligent drilling monitor[R]. SPE 59227, 2000.
[23] ZHU J, ZENG L. Intelligent pressure control system on drilling process[R]. OTC 30828, 2020.
[24] SHEN Xinyu, LIU Sen, SU Qiang, et al. Intelligent switch control algorithm of the push-the-bit rotary steerable drilling system[R]. ARMA 20233-0558, 2023.
[25] ROWSELL P J, WALLER M D. Intelligent control of drilling systems[R]. SPE 21927, 1991.
[26] WAN Youwei, LIU Xiangjun, XIONG Jian, et al. Intelligent prediction of drilling rate of penetration based on method-data dual validity analysis[J]. SPE Journal, 2024, 29(5): 2257–2274. doi: 10.2118/217977-PA
[27] ABUGHABAN M, ALSHAARAWI A, MENG Cui, et al. Optimization of drilling performance based on an intelligent drilling advisory system[R]. IPTC 19269, 2019.
[28] XIE Tao, HOU Xinxin, HUO Hongbo, et al. Improving drilling efficiency using intelligent decision system for drilling in Bohai Oilfield based on big data[R]. SPE 215427, 2023.
[29] RASHIDI B, HARELAND G, TAHMEEN M, et al. Real-time bit wear optimization using the intelligent drilling advisory system[R]. SPE 136006, 2010.
[30] WANG Jianhua, GUAN Zhen, LIU Muchen, et al. Drilling stuck probability intelligent prediction based on LSTM considering local interpretability[R]. ARMA 2023-0326, 2023.
[31] 殷启帅,杨进,曹博涵,等. 基于长短期记忆神经网络的深水钻井工况实时智能判别模型[J]. 石油钻采工艺,2022,44(1):97–104. YIN Qishuai, YANG Jin, CAO Bohan, et al. Real-time intelligent rig activities classification model of deep-water drilling using long short-term memory (LSTM) network[J]. Oil Drilling & Production Technology, 2022, 44(1): 97–104.
[32] 李雪松,张骁,管震,等. 基于图像识别技术的钻井井漏溢流智能报警系统开发[J]. 世界石油工业,2021,28(1):48–54. LI Xuesong, ZHANG Xiao, GUAN Zhen, et al. Development of the drilling mud loss and overflow intelligent alarm system based on the image recognition technology[J]. World Petroleum Industry, 2021, 28(1): 48–54.
[33] WANG Han, CHEN Dong, YE Zhihui, et al. Intelligent planning of drilling trajectory based on computer vision[R]. SPE 197362, 2019.
[34] 张晓东,朱正喜. 智能钻井技术研究[J]. 石油钻采工艺,2010,32(1):1–4. doi: 10.3969/j.issn.1000-7393.2010.01.002 ZHANG Xiaodong, ZHU Zhengxi. Study of intelligent drilling technology[J]. Oil Drilling & Production Technology, 2010, 32(1): 1–4. doi: 10.3969/j.issn.1000-7393.2010.01.002
[35] de WARDT J. Guest editorial: trends in remote operations and drilling systems automation point to an expanding footprint what comes next and when?[J]. Journal of Petroleum Technology, 2022, 74(11): 10–13. doi: 10.2118/1122-0010-JPT
[36] 刘合. 油气勘探开发数字化转型人工智能应用大势所趋[J]. 石油科技论坛,2023,42(3):1–9. LIU He. Digital transformation of oil and gas exploration and development; unstoppable AI application[J]. Petroleum Science and Technology Forum, 2023, 42(3): 1–9.
[37] 宋先知,姚学喆,李根生,等. 基于LSTM-BP神经网络的地层孔隙压力计算方法[J]. 石油科学通报,2022,7(1):12–23. doi: 10.3969/j.issn.2096-1693.2022.01.002 SONG Xianzhi, YAO Xuezhe, LI Gensheng, et al. A novel method to calculate formation pressure based on the LSTM-BP neural network[J]. Petroleum Science Bulletin, 2022, 7(1): 12–23. doi: 10.3969/j.issn.2096-1693.2022.01.002
[38] ROWE H, MAINALI P, NIETO M, et al. Geochemical perspectives on cuttings-based chemostratigraphy and mineral modeling in the Delaware Basin, Texas and New Mexico[R]. URTEC 2019-1068, 2019.
[39] HUSSAIN M, AMAO A, AL-RAMADAN K, et al. A novel method to develop chemostratigraphy using X-ray fluorescence spectral raw data[R]. URTEC 2021-5478, 2021.
[40] MICHAEL N A, SCHEIBE C, CRAIGIE N W. Automations in chemostratigraphy: toward robust chemical data analysis and interpretation[R]. SPE 204892, 2021.
[41] 唐诚,王崇敬,梁波,等. 基于机器学习算法的页岩气评价参数计算模型研究[J]. 录井工程,2021,32(4):18–22. doi: 10.3969/j.issn.1672-9803.2021.04.003 TANG Cheng, WANG Chongjing, LIANG Bo, et al. Study on shale gas evaluation parameter calculation model based on machine learning algorithm[J]. Mud Logging Engineering, 2021, 32(4): 18–22. doi: 10.3969/j.issn.1672-9803.2021.04.003
[42] 刘雨龙. 基于深度学习的岩屑智能分析方法研究[D]. 北京:中国石油大学(北京),2023. LIU Yulong. Research on intelligent analysis method of cuttings based on deep learning[D]. Beijing: China University of Petro-leum(Beijing), 2023.
[43] 夏文鹤,谢万洋,唐印东,等. 砂样岩屑图像特征的岩性智能高效识别[J]. 石油地球物理勘探,2023,58(3):495–506. XIA Wenhe, XIE Wanyang, TANG Yindong, et al. Intelligent and efficient lithology identification based on image features of returned cuttings[J]. Oil Geophysical Prospecting, 2023, 58(3): 495–506.
[44] 张德君,魏伟,刘明艳,等. 基于综合录井数据的地层岩性智能识别方法[J]. 西部探矿工程,2023,35(4):54–59. doi: 10.3969/j.issn.1004-5716.2023.04.016 ZHANG Dejun, WEI Wei, LIU Mingyan, et al. Intelligent identification method of formation lithology based on comprehensive logging data[J]. West-China Exploration Engineering, 2023, 35(4): 54–59. doi: 10.3969/j.issn.1004-5716.2023.04.016
[45] JACOBS T. Mud-gas breakthrough Equinor develops real-time reservoir-fluid identification[J]. Journal of Petroleum Technolog, 2021, 73(2): 37–39. doi: 10.2118/0221-0037-JPT
[46] HAFIDZ ARIEF I, YANG Tao. Real time reservoir fluid log from advanced mud gas data[R]. SPE 201323, 2020.
[47] YANG Tao, YERKINKYZY G, ULEBERG K, et al. Predicting reservoir fluid properties from advanced mud gas data[J]. SPE Reservoir Evaluation & Engineering, 2021, 24(2): 358–366.
[48] UNGAR F, MCGILL A, NYGAARD M T, et al. Fluid identification from mud gas in the overburden: a case study for the Snorre Field[R]. SPE 214440, 2023.
[49] YANG Tao, ULEBERG K, CELY A, et al. Unlock large potentials of standard mud gas for real-time fluid typing[R]. SPWLA 2022-0007, 2022.
[50] KOPAL M, YERKINKYZY G, NYGÅRD M T, et al. Real-time fluid identification from integrating advanced mud gas and petrophysical logs[R]. SPWLA 2022-0009, 2022.
[51] BUCKLE P S G, ABDULLAH A F H, ZAINI N, et al. Utilization of digitalized numerical model derived from advanced mud gas data for low cost fluid phase identification, derisking drilling and effective completion plan in depleted reservoir[R]. SPWLA 2022-0092, 2022.
[52] WRIGHT A C. Estimation of gas/oil ratios and detection of unusual formation fluids from mud logging gas data[R]. SPWLA 1996-CC, 1996.
[53] MALIK M, HANSON S A, CLINCH S. Maximizing value from mudlogs: integrated approach to determine net pay[R]. SPWLA 5028, 2020.
[54] 严伟丽,高楚桥,赵彬,等. 基于气测录井资料的气油比定量计算方法[J]. 科学技术与工程,2020,20(23):9287–9292. YAN Weili, GAO Chuqiao, ZHAO Bin, et al. Quantitative calculation method of gas-oil ratio in gas logging data[J]. Science Technology and Engineering, 2020, 20(23): 9287–9292.
[55] JIANG Han, DAIGLE H, TIAN Xiao, et al. A comparison of clustering algorithms applied to fluid characterization using NMRT1-T2 maps of shale[J]. Computers & Geosciences, 2019, 126: 52–61.
[56] 夏文鹤,唐印东,李皋,等. 基于岩屑录井图像的井壁稳定性智能预测方法[J]. 天然气工业,2023,43(12):71–83. XIA Wenhe, TANG Yindong, LI Gao, et al. An intelligent prediction method for wellbore stability based on drilling cuttings logging images[J]. Natural Gas Industry, 2023, 43(12): 71–83.
[57] ZHANG Shaohui, HUANG Weihe, BI Guoqiang, et al. Intelligent risk identification and warning model for typical drilling operation scenes and its application[R]. SPE 214599, 2023.
[58] 胡志强,杨进,王磊,等. 钻井工况智能识别与时效分析技术[J]. 石油钻采工艺,2022,44(2):241–246. HU Zhiqiang, YANG Jin, WANG Lei, et al. Intelligent identification and time-efficiency analysis of drilling operation conditions[J]. Oil Drilling & Production Technology, 2022, 44(2): 241–246.
[59] 张矿生,宫臣兴,陆红军,等. 基于集成学习的井漏智能预警模型及智能推理方法[J]. 石油钻采工艺,2023,45(1):47–54. ZHANG Kuangsheng, GONG Chenxing, LU Hongjun, et al. Intelligent early warning model and intelligent reasoning method based on integrated learning for loss circulation[J]. Oil Drilling & Production Technology, 2023, 45(1): 47–54.
[60] 张敏. 渤海中深层钻井地层三压力智能预测方法研究[D]. 北京:中国石油大学(北京),2023. ZHANG Min. Research on the intelligent prediction method of triple pressure in the middle and deep drilling strata in Bohai Sea[D]. Beijing: China University of Petroleum(Beijing), 2023.
[61] 阎荣辉,黄子舰,杨永强,等. “互联网+” 时代的智慧录井系统应用探索[J]. 录井工程,2020,31(2):1–5. doi: 10.3969/j.issn.1672-9803.2020.02.001 YAN Ronghui, HUANG Zijian, YANG Yongqiang, et al. Application and exploration of smart logging system in the Internet Plus era[J]. Mud Logging Engineering, 2020, 31(2): 1–5. doi: 10.3969/j.issn.1672-9803.2020.02.001
[62] 方铁园,马宏伟,郭龙飞,等. 智慧录井平台建设及其在长庆油田的创新应用[J]. 录井工程,2023,34(3):97–101. doi: 10.3969/j.issn.1672-9803.2023.03.015 FANG Tieyuan, MA Hongwei, GUO Longfei, et al. Construction of smart mud logging platform and its innovative application in Changqing Oilfield[J]. Mud Logging Engineering, 2023, 34(3): 97–101. doi: 10.3969/j.issn.1672-9803.2023.03.015
[63] 张锦宏,周爱照,成海,等. 中国石化石油工程技术新进展与展望[J]. 石油钻探技术,2023,51(4):149–158. ZHANG Jinhong, ZHOU Aizhao, CHENG Hai, et al. New progress and prospects for Sinopec’s petroleum engineering technologies[J]. Petroleum Drilling Techniques, 2023, 51(4): 149–158.
[64] 梁海波,宋洋,于志刚,等. 钻井液流变性实时测量方法及系统研究[J]. 石油机械,2022,50(1):10–18. LIANG Haibo, SONG Yang, YU Zhigang, et al. Real-time measurement method and systematic study on drilling fluid rheology[J]. China Petroleum Machinery, 2022, 50(1): 10–18.
[65] 孟济良,吴龙斌,张学亭. 岩屑录井的新技术:DL-1型地质录井自动捞砂机介绍[J]. 石油勘探与开发,1984,11(5):75–79. MENG Jiliang, WU Longbin, ZHANG Xueting. A new technology of cuttings logging: DL-1 type automatic cuttings acquisition machine[J]. Petroleum Exploration and Development, 1984, 11(5): 75–79.
[66] ALSHEHRI A, KATTERBAUER K, YOUSEF A. Real-time autoregressive deep learning framework for in-line automatic surface logging[R]. SPE 214079, 2023.
[67] LI Sanguo, XIAO Lizhi, LI Xin, et al. A novel NMR instrument for real time drilling fluid analysis[J]. Microporous and Mesoporous Materials, 2018, 269: 138–141. doi: 10.1016/j.micromeso.2017.08.038
[68] 张志财,刘保双,王忠杰,等. 钻井液性能在线监测系统的研制与现场应用[J]. 钻井液与完井液,2020,37(5):597–601. ZHANG Zhicai, LIU Baoshuang, WANG Zhongjie, et al. Development and field application of an online drilling fluid property monitoring system[J]. Drilling Fluid & Completion Fluid, 2020, 37(5): 597–601.
[69] 王鹏,刘伟,张果. 钻井液性能自动监测装置的现状及改进建议[J]. 钻采工艺,2022,45(3):42–47. doi: 10.3969/J.ISSN.1006-768X.2022.03.08 WANG Peng, LIU Wei, ZHANG Guo. Status quo and improvement suggestions of automatic monitoring equipment for drilling fluid performance[J]. Drilling & Production Technology, 2022, 45(3): 42–47. doi: 10.3969/J.ISSN.1006-768X.2022.03.08
[70] 张好林,杨传书,李昌盛,等. 钻井数字孪生系统设计与研发实践[J]. 石油钻探技术,2023,51(3):58–65. ZHANG Haolin, YANG Chuanshu, LI Changsheng, et al. Design and research practice of a drilling digital twin system[J]. Petroleum Drilling Techniques, 2023, 51(3): 58–65.
[71] BALACHANDRAN P A, PADMANABHAN K V K. Integrated operations system: implementation of a truly integrated digital oil field and development of digital twin[R]. OTC 32841, 2023.
[72] WANG Peixian, WANG Xiaolin, MA Jun, et al. Digital and intelligent technology in underground gas storage operation based on digital twin technologies[R]. ISOPE I-23-008, 2023.
[73] BUSOLLO C, ABDO E, KATTAR M, et al. Evolution of a digital twin for underground gas storage wells: thermal effects of tubing gas flow on annulus pressure in transient conditions[R]. SPE 220006, 2024.
[74] WANG Peixian, MA Jun, LI Zunzhao, et al. Integrated simulation technology of underground gas storage based on digital twin technologies[R]. ISOPE I-24-043, 2024.
[75] 刘合,任义丽,李欣,等. 油气行业人工智能大模型应用研究现状及展望[J]. 石油勘探与开发,2024,51(4):910–923. LIU He, REN Yili, LI Xin, et al. Research status and application of artificial intelligence large models in the oil and gas industry[J]. Petroleum Exploration and Development, 2024, 51(4): 910–923.
[76] YI M, CEGLINSKI K, ASHOK P, et al. Applications of large language models in well construction planning and real-time operation[R]. SPE 217700, 2024.
[77] PACIS F J, ALYAEV S, PELFRENE G, et al. Enhancing information retrieval in the drilling domain: zero-shot learning with large language models for question-answering[R]. SPE 217671, 2024.
[78] AMEUR-ZAIMECHE O, KECHICHED R, HEDDAM S, et al. Real-time porosity prediction using gas-while-drilling data and machine learning with reservoir associated gas: case study for Hassi Messaoud Field, Algeria[J]. Marine and Petroleum Geology, 2022, 140: 105631. doi: 10.1016/j.marpetgeo.2022.105631
[79] OULADMANSOUR A, AMEUR-ZAIMECHE O, KECHICHED R, et al. Integrating drilling parameters and machine learning tools to improve real-time porosity prediction of multi-zone reservoirs. Case study: Rhourd Chegga Oilfield, Algeria[J]. Geoenergy Science and Engineering, 2023, 223: 211511.
计量
- 文章访问数: 166
- HTML全文浏览量: 30
- PDF下载量: 98
