油气开发

基于改进LSTM神经网络的加密井产能预测研究——以川南中深层页岩气为例

  • 官文洁 ,
  • 彭小龙 ,
  • 朱苏阳 ,
  • 杨晨 ,
  • 彭真 ,
  • 马潇然
展开
  • 西南石油大学油气藏地质及开发工程全国重点实验室,四川 成都 610500
官文洁(2000—),女,在读硕士研究生,主要从事石油领域人工智能算法研究。地址:四川省成都市新都区新都大道8号,邮政编码:610500。E-mail:13628038631@163.com
彭小龙(1973—),男,博士,教授,博士生导师,主要从事油气及煤层气藏渗流理论和数值模拟、地质建模数模一体化和油气藏分子模拟研究。地址:四川省成都市新都区新都大道8号,邮政编码:610500。E-mail:peng_xl@126.com

收稿日期: 2024-06-27

  网络出版日期: 2025-05-08

基金资助

国家自然科学基金项目“基于煤粉群运移动力学特征的煤层气-水-固耦合传质机理研究”(52104036);四川省自然科学基金项目“考虑纳米尺度效应与微纳跨尺度流动的煤层气临界解吸机制研究”(2023NSFSC0932)

Research on productivity prediction method of infilling well based on improved LSTM neural network: A case study of the middle-deep shale gas in South Sichuan

  • GUAN Wenjie ,
  • PENG Xiaolong ,
  • ZHU Suyang ,
  • YANG Chen ,
  • PENG Zhen ,
  • MA Xiaoran
Expand
  • State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation, Southwest Petroleum University, Chengdu, Sichuan 610500, China

Received date: 2024-06-27

  Online published: 2025-05-08

摘要

川南中深层页岩气开发过程中,常规油气藏工程方法,如裂缝扩展、应力诱导分析和数值模拟等研究过程使得加密井的预测工作繁重,且无法有效应对不同生产阶段的产能差异性,应用条件苛刻。为了快速且准确预测加密井产能,根据老井生产压力曲线呈趋势性“三段式”递减的特征,将剧烈下降期作为前期产水期,快速下降和缓慢下降期作为后期产气期两部分,采用优化速度快、具有自适应性和信息反馈机制的灰狼优化算法(GWO)对长短期记忆(LSTM)神经网络模型进行超参数择优,分别构建由GWO计算最优解确定隐含层神经元个数、丢包率和批次数的前、后期模型,通过损失曲线和性能指标曲线确定迭代次数,采用线性学习率热身的方法动态调整学习率,实现高速训练过程,形成分阶段的产量预测模型。实例研究表明:GWO优化的LSTM神经网络模型在预设学习率为0.002、迭代450次的条件下,短时间内能够快速实现收敛,最终性能指标达到0.923。GWO优化的LSTM神经网络模型与传统LSTM神经网络模型预测结果相比,前、后期平均绝对误差分别降低了1.290 m3/d和0.213×104 m3/d;与数值模拟拟合结果相比,产气量预测的平均绝对误差降低了0.24×104 m3/d。因此,改进后的LSTM神经网络模型在不同生产阶段的产能预测中表现出色,且对应阶段模型能够准确预测川南中深层页岩气加密井的产能变化,为加密井产能预测方法提供理论依据。

本文引用格式

官文洁 , 彭小龙 , 朱苏阳 , 杨晨 , 彭真 , 马潇然 . 基于改进LSTM神经网络的加密井产能预测研究——以川南中深层页岩气为例[J]. 油气藏评价与开发, 2025 , 15(3) : 479 -487 . DOI: 10.13809/j.cnki.cn32-1825/te.2025.03.015

Abstract

During the development of middle and deep gas reservoirs in South Sichuan, conventional reservoir engineering methods—such as fracture propagation, stress-induced analysis, and numerical simulation—render productivity prediction of infilling wells laborious and ineffective in addressing variations in production capacity across different production stages, with stringent application conditions. In order to quickly and accurately predict the production capacity of infilling wells, this study classifies the “three-stage” declining trend observed in the production pressure curves of existing wells into: (1) A drastic decline period, regarded as the initial water production stage; (2) a rapid decline period; and (3) a slow decline period, both considered part of the later gas production stage. The Grey Wolf Optimizer(GWO) algorithm, a fast optimization algorithm with adaptive capabilities and an information feedback mechanism, is applied for hyperparameter optimization of the Long Short-term Memory (LSTM) neural network. Two stage-specific models were constructed, with the number of hidden layer neurons, dropout rate, and batch size determined by the optimal solutions obtained via GWO. The number of iterations was selected based on the loss curve and performance metric curve, while a linear warm-up strategy was used to dynamically adjust the learning rate, facilitating high-speed training and the formation of a staged productivity prediction model. Example studies show that the GWO-optimised LSTM neural network model achieves rapid convergence with a preset learning rate of 0.002 and 450 iterations, ultimately reaching a performance index of 0.923. Compared to the conventional LSTM neural network model, the average absolute errors during the early and later stages are reduced by 1.290 m3/d and 0.213 × 104 m3/d, respectively. Compared with numerical simulation fitting results, the average absolute error in gas production prediction is reduced by 0.24 × 104 m3/d. Therefore, the improved LSTM neural network model demonstrates excellent performance in capacity prediction across different production stages, and the stage-specific productivity variations in infilling wells within middle and deep shale gas reservoirs in South Sichuan. This provides a theoretical foundation for productivity prediction methods of infilling wells.

参考文献

1 许宁, 张俊杰, 叶锋, 等. 非常规油藏蓄能体积压裂的合理压裂规模研究[J]. 石油地质与工程, 2024, 38(2): 97-101.
  XU Ning, ZHANG Junjie, YE Feng, et al. Study on reasonable fracturing scale of accumulative volume fracturing in unconventional reservoir[J]. Petroleum Geology and Engineering, 2024, 38(2): 97-101.
2 王世禄. 松辽盆地古龙页岩油水平井压裂施工参数优化[J]. 石油地质与工程, 2023, 37(5): 94-99.
  WANG Shilu. Optimization of fracturing construction parameters for Gulong shale oil with horizontal wells in Songliao Basin[J]. Petroleum Geology and Engineering, 2023, 37(5): 94-99.
3 苏玉亮, 盛广龙, 王文东, 等. 页岩气藏多重介质流体跨尺度传质表征方法研究[J]. 中国科学: 技术科学, 2018, 48(5): 510-523.
  SU Yuliang, SHENG Guanglong, WANG Wendong, et al. Characterization methods for mass transfer of multiple media in shale gas reservoirs[J]. Scientia Sinica(Technologica), 2018, 48(5): 510-523.
4 刘子雄, 李敬松, 黄子俊, 等. 利用录井资料计算低渗气井压裂产能的新方法[J]. 断块油气田, 2014, 21(4): 497-499.
  LIU Zixiong, LI Jingsong, HUANG Zijun, et al. A new method to calculate fracturing productivity of low permeability gas well using logging data[J]. Fault-Block Oil and Gas Field, 2014, 21(4): 497-499.
5 赵金洲, 李志强, 胡永全, 等. 考虑页岩储层微观渗流的压裂产能数值模拟[J]. 天然气工业, 2015, 35(6): 53-58.
  ZHAO Jinzhou, LI Zhiqiang, HU Yongquan, et al. Numerical simulation of productivity after fracturing with consideration to micro-seepage in shale reservoirs[J]. Natural Gas Industry, 2015, 35(6): 53-58.
6 李泽沛, 彭小龙, 王毅. 基于三重介质模型的体积压裂后页岩气储层数值模拟方法[J]. 油气地质与采收率, 2016, 23(6): 105-111.
  LI Zepei, PENG Xiaolong, WANG Yi. Numerical simulation method of shale gas reservoirs after stimulated reservoir volume fracturing based on triple porous media model[J]. Petroleum Geology and Recovery Efficiency, 2016, 23(6): 105-111.
7 张晨招. 新的页岩气多级压裂水平井产能预测模型研究[J]. 地下水, 2018, 40(5): 121-125.
  ZHANG Chenzhao. New shale gas multi-stage fracturing horizontal well productivity prediction model research[J]. Ground Water, 2018, 40(5): 121-125.
8 张春晓, 马含含, 侯梦瑶, 等. 基于时间序列分析的油井产量预测应用综述[C]// 2023国际石油石化技术会议论文集. 西安: 西安石油大学, 2023.
  ZHANG Chunxiao, MA Hanhan, HOU Mengyao, et al. A review on application of oil well production prediction based on time series analysis method[C]// Proceedings of 2023 International Petroleum and Petrochemical Technology Conference. Xi’an: Xi’an Petroleum University, 2023.
9 强璐, 任宇飞, 马焕焕, 等. 基于GA-BP神经网络的特低渗砂岩油藏水平井产能预测[J]. 石油地质与工程, 2024, 38(6): 63-67.
  QIANG Lu, REN Yufei, MA Huanhuan, et al. Prediction of horizontal well productivity in ultra-low permeability sandstone reservoirs based on GA-BP neutral network[J]. Petroleum Geology & Engineering, 2024, 38(6): 63-67.
10 赵林丰, 李晓静, 王晶晶, 等. 多属性神经网络反演在重力流储层预测中的应用: 以歧口凹陷歧南斜坡沙一段为例[J]. 石油地质与工程, 2024, 38(5): 7-12.
  ZHAO Linfeng, LI Xiaojing, WANG Jingjing, et al. Application of multi-attribute neural network inversion in gravity flow reservoir prediction: A case study of Es1 Formation in Qinan Slope, Qikou Sag[J]. Petroleum Geology & Engineering, 2024, 38(5): 7-12.
11 田亚鹏, 鞠斌山. 基于遗传算法改进BP神经网络的页岩气产量递减预测模型[J]. 中国科技论文, 2016, 11(15): 1710-1715.
  TIAN Yapeng, JU Binshan. A model for predicting shale gas production decline based on the BP neural network improved by the genetic algorithm[J]. China Sciencepaper, 2016, 11(15): 1710-1715.
12 赵军, 夏宏泉, 刘红歧. 基于BP神经网络的油气产量历史预测[J]. 西南石油学院学报, 1998, 20(2): 23-26.
  ZHAO Jun, XIA Hongquan, LIU Hongqi. Prediction of oil/gas history based on neural network[J]. Journal of Southwest Petroleum Institute, 1998, 20(2): 23-26.
13 周于皓, 刘慧卿, 祁鹏, 等. 基于循环神经网络的缝洞型油藏油井产量预测[J]. 计算物理, 2018, 35(6): 668-674.
  ZHOU Yuhao, LIU Huiqing, QI Peng, et al. Forecast of oil production in fractured-vuggy reservoir by using recurrent neural networks[J]. Chinese Journal of Computational Physics, 2018, 35(6): 668-674.
14 高亚军, 唐力辉, 王振鹏, 等. 基于循环神经网络和数据差分处理的油田产量预测方法[J]. 中国海上油气, 2023, 35(3): 126-136.
  GAO Yajun, TANG Lihui, WANG Zhenpeng, et al. A method for oilfield production prediction based on recurrent neural network and data differential processing[J]. China Offshore Oil and Gas, 2023, 35(3): 126-136.
15 张春晓. 基于深度长短期记忆神经网络的油气井产量预测优化方法[J]. 石油化工应用, 2023, 42(11): 28-31.
  ZHANG Chunxiao. Optimization method for oil and gas well production prediction based on deep long short-term memory neural network[J]. Petrochemical Industry Application, 2023, 42(11): 28-31.
16 王洪亮, 林霞, 蒋丽维, 等. 基于聚类及长短时记忆神经网络预测油田产量[J]. 石油科学通报, 2024, 9(1): 62-72.
  WANG Hongliang, LIN Xia, JIANG Liwei, et al. An oilfield production prediction method based on clustering and long short-term memory neural network[J]. Petroleum Science Bulletin, 2024, 9(1): 62-72.
17 刘嘉豪, 刘浩. 基于LSTM神经网络模型的石油单井产量预测[J]. 石油化工应用, 2023, 42(3): 38-41.
  LIU Jiahao, LIU Hao. Production prediction of single oil well based on LSTM neural network model[J]. Petrochemical Industry Application, 2023, 42(3): 38-41.
18 韩克宁, 王伟, 樊冬艳, 等. 基于产量递减与LSTM耦合的常压页岩气井产量预测[J]. 油气藏评价与开发, 2023, 13(5): 647-656.
  HAN Kening, WANG Wei, FAN Dongyan, et al. Production forecasting for normal pressure shale gas wells based on coupling of production decline method and LSTM model[J]. Petroleum Reservoir Evaluation and Development, 2023, 13(5): 647-656.
19 付钰绮, 王杨, 吴思樵, 等. 基于CNN-LSTM-ATT网络的页岩气井产量预测[J]. 天然气技术与经济, 2024, 18(2): 32-38.
  FU Yuqi, WANG Yang, WU Siqiao, et al. Predicting the production in shale gas wells based on CNN-LSTM-ATT model[J]. Natural Gas Technology and Economy, 2024, 18(2): 32-38.
20 缪欢, 郑洪扬, 范文龙, 等. 四川盆地龙马溪组深层页岩储层压力与含气量动态演化过程[J]. 世界石油工业, 2024, 31(5): 19-29, 39.
  MIAO Huan, ZHENG Hongyang, FAN Wenlong, et al. Dynamic evolution process of pressure and gas content in the Longmaxi Formation deep shale reservoir of Sichuan Basin[J]. World Petroleum Industry, 2024, 31(5): 19-29.
21 史伟光, 王启任. 基于改进灰狼算法的多任务优化算法[J]. 天津工业大学学报, 2023, 42(5): 81-86.
  SHI Weiguang, WANG Qiren. Improved grey wolf algorithm based multitask optimization algorithm[J]. Journal of Tiangong University, 2023, 42(5): 81-86.
22 秦宏伍, 王立铮, 傅渝, 等. 基于多策略结合的灰狼优化算法及应用[J]. 山东大学学报(理学版), 2024, 59(3): 51-60.
  QIN Hongwu, WANG Lizheng, FU Yu, et al. Grey wolf optimization algorithm based on multi-strategy combination and its application[J]. Journal of Shandong University (Natural Science), 2024, 59(3): 51-60.
23 姜云. 灰狼优化混合算法及其K-means聚类优化研究[D]. 新乡: 河南师范大学, 2021.
  JIANG Yun. Research on hybrid grey wolf optimizers and their K-means clustering optimization[D]. Xinxiang: Henan Normal University, 2021.
24 夏雪飞. 焦石坝背斜中北部页岩气井压力变化规律研究[J]. 江汉石油职工大学学报, 2018, 31(2): 14-17.
  XIA Xuefei. Research on pressure change pattern of north central shale gas wells in jiaoshiba anticline[J]. Journal of Jianghan Petroleum University of Staff and Workers, 2018, 31(2): 14-17.
25 商晓飞, 段太忠, 包汉勇, 等. 基于裂缝相表征的页岩气藏天然裂缝新模型:以涪陵页岩气田焦石坝区块为例[J]. 天然气工业, 2023, 43(6): 44-56.
  SHANG Xiaofei, DUAN Taizhong, BAO Hanyong, et al. A new model of natural fractures in shale gas reservoirs based on fracture facies characterization: A case study from the Jiaoshiba Block of the Fuling Shale Gas Field[J]. Natural Gas Industry, 2023, 43(6): 44-56.
26 贾承造, 王祖纲, 姜林, 等.中国油气勘探开发成就与未来潜力: 深层、深水与非常规油气: 专访中国科学院院士、石油地质与构造地质学家贾承造[J].世界石油工业, 2023, 30(3): 1-8.
  JIA Chengzao, WANG Zugang, JIANG Lin, et al. Achievements and future potential for oil&gas exploration and development in China: Deep-formation, deep-water and unconventional reservoirs: Interview with JIA Chengzao, Academician of the CAS, geologist in petroleum geology and structure[J]. World Petroleum Industry, 2023, 30(3): 1-8.
27 汪泽成, 赵振宇, 黄福喜, 等. 中国中西部含油气盆地超深层油气成藏条件与勘探潜力分析[J]. 世界石油工业, 2024, 31(1): 33-48.
  WANG Zecheng, ZHAO Zhenyu, HUANG Fuxi, et al. Ultra-deep hydrocarbon accumulation conditions and exploration potential in sedimentary in sedimentary basins of Central-Western China[J]. World Petroleum Industry, 2024, 31(1): 33-48.
28 刘方圆. 涪陵页岩气田焦页XHF井生产动态分析及建议[J]. 地质科技情报, 2018, 37(3): 196-201.
  LIU Fangyuan. Advices for Well Jiaoye XHF, Fuling Shale gas field by analyzing production data[J]. Geological Science and Technology Information, 2018, 37(3): 196-201.
文章导航

/