基于改进人工神经网络的页岩气井产量预测模型研究

doi:10.13809/j.cnki.cn32-1825/te.2023.04.008

摘要/Abstract

摘要：

传统页岩气井产量预测方法难以对储层参数、压裂参数与产量的关系做出有效分析，而机器学习方法具有解决这一问题的能力。提出了基于物理意义和随机组合的方法构建特征参数，并采用小批量梯度下降法（MBGD）作为训练函数，建立了针对页岩气井产量预测的改进人工神经网络预测模型。然后结合实例，利用改进后的人工神经网络模型对页岩气井产量进行预测，并通过计算均方误差（MSE）和修正决定系数（T）的值对模型的优劣程度和预测精度进行评价。结果表明，建立的改进神经网络模型预测产量结果与实际产量值吻合度较高，且相比传统的BP（误差反向传播算法）神经网络模型，在预测精度和稳定性方面具有明显优势。该模型能为页岩气储层压裂优化设计以及产能评价提供重要支持。

关键词: 人工神经网络, 页岩气, 压裂参数, 产量预测, 特征构建

Abstract:

Traditional methods for predicting shale gas well production often struggle to effectively analyze the complex relationship between reservoir parameters, fracturing parameters and production. To address these challenges, a novel approach is introduced, involving the construction of characteristic parameters based on physical meaning and random combination. The small batch gradient descent method（MBGD） is adopted as the training function to develop an improved artificial neural network prediction model for shale gas well production. An example is utilized to demonstrate the effectiveness of the improved artificial neural network model in predicting shale gas well production. The model’s performance is evaluated using the mean squared error（MSE） and the modified determination coefficient（T）. The results indicate that the predictions from the improved network model align well with the actual production data. Moreover, the model exhibits superior prediction accuracy and stability compared to the traditional BP（error backpropagation algorithm） neural network model. With its high accuracy and reliability, the proposed model can provide valuable support for fracturing optimization design and productivity evaluation in shale gas reservoirs.

Key words: artificial neural network, shale gas, fracturing parameter, production forecast, feature construction

中图分类号:

TE32

林魂, 孙新毅, 宋西翔, 蒙春, 熊雯欣, 黄俊和, 刘洪博, 刘成. 基于改进人工神经网络的页岩气井产量预测模型研究[J]. 油气藏评价与开发, 2023, 13(4): 467-473.

LIN Hun, SUN Xinyi, SONG Xixiang, MENG Chun, XIONG Wenxin, HUANG Junhe, LIU Hongbo, LIU Cheng. A model for shale gas well production prediction based on improved artificial neural network[J]. Petroleum Reservoir Evaluation and Development, 2023, 13(4): 467-473.

图/表 10

表1

表2

表3

图1

图2

图3

图4

表4

图5

表5

参考文献 22

[1]	何叶, 张涵冰, 郑儒, 等. 基于元素录井的页岩气水平井钻遇小层分析及储层评价参数计算[J]. 天然气工业, 2021, 41(增刊1): 110-117.
	HE Ye, ZHANG Hanbing, ZHENG Ru, et al. Analyzing the sublayers drilled by shale-gas horizontal wells and calculating reservoir evaluation parameters based on element logging[J]. Natural Gas Industry, 2021, 41 (sup.1): 110-117.
[2]	王燕, 雷有为, 付小平, 等. 涪陵区块凉高山组页岩气储层特征及关键参数评价[J]. 复杂油气藏, 2020, 13(4): 23-28.
	WANG Yan, LEI Youwei, FU Xiaoping, et al. Characteristics and key parameter evaluation of shale gas reservoirsin Lianggaoshan Formation in Fuling Block[J]. Complex Hydrocarbon Reservoirs, 2020, 13(4): 23-28.
[3]	李亚龙, 刘先贵, 胡志明, 等. 页岩气水平井产能预测数值模型综述[J]. 地球科学进展, 2020, 35(4): 350-362. doi: 10.11867/j.issn.1001-8166.2020.037
	LI Yalong, LIU Xiangui, HU Zhiming, et al. Summary of numerical models for predicting productivity of shale gas horizontal wells[J]. Advances in Earth Science, 2020, 35(4): 350-362. doi: 10.11867/j.issn.1001-8166.2020.037
[4]	陈元千, 徐佳倩, 傅礼兵. 预测页岩气井产量和可采储量泛指数递减模型的建立及应用[J]. 油气地质与采收率, 2021, 28(1): 132-136.
	CHEN Yuanqian, XU Jiaqian, FU Libing. Establishment and application of pan exponential decline model for forecasting production rate and recoverable reserves of shale gas wells[J]. Petroleum Geology and Recovery Efficiency, 2021, 28(1): 132-136.
[5]	王清媛, 黄全舟. 浅析机器学习在石油测井领域的研究进展[J]. 清洗世界, 2021, 37(3): 120-122.
	WANG Qingyuan, HUANG Quanzhou. Research progress of machine learning in oil logging[J]. Cleaning world, 2021, 37(3): 120-122.
[6]	闵超, 代博仁, 张馨慧, 等. 机器学习在油气行业中的应用进展综述[J]. 西南石油大学学报(自然科学版), 2020, 42(6): 1-15. doi: 10.11885/j.issn.1674-5086.2020.06.05.03
	MIN Chao, DAI Boren, ZHANG Xinhui, et al. A review of the application progress of machine learning in oil and gas industry[J]. Journal of Southwest Petroleum University (Science & Technology Edition), 2020, 42(6): 1-15. doi: 10.11885/j.issn.1674-5086.2020.06.05.03
[7]	NEGASH Berihun Mamo, YAW Atta Dennis. 基于人工神经网络的注水开发油藏产量预测[J]. 石油勘探与开发, 2020, 47(2): 357-365. doi: 10.11698/PED.2020.02.14
	NEGASH Berihun Mamo, YAW Atta Dennis. Artificial neural network based production forecasting for a hydrocarbon reservoir under water injection[J]. Petroleum Exploration and Development, 2020, 47(2): 357-365. doi: 10.11698/PED.2020.02.14
[8]	林年添, 张栋, 张凯, 等. 地震油气储层的小样本卷积神经网络学习与预测[J]. 地球物理学报, 2018, 61(10): 4110-4125.
	LIN NianTian, ZHANG Dong, ZHANG Kai, et al. Predicting distribution of hydrocarbon reservoirs with seismic data based on learning of the small-sample convolution neural network[J]. Chinese Journal of Geophysics, 2018, 61(10): 4110-4125.
[9]	ZHANG H Q, YU F S, SUN J C, et al. Deep learning for sea cucumber detection using stochastic gradient descent algorithm[J]. European Journal of Remote Sensing, 2020, 53 (sup.1): 53-62.
[10]	李英, 贺春林. 面向深度神经网络训练的数据差分隐私保护随机梯度下降算法[J]. 计算机应用与软件, 2020, 37(4): 252-259.
	LI Ying, HE Chunlin. Data differential privacy protection stochastic gradient descent algorithm for deep neural network training[J]. Computer Applications and Software, 2020, 37(4): 252-259.
[11]	穆翔宇, 范钰, 李苏吉, 等. 一种基于梯度下降算法的蜕变关系生成方法[J]. 吉林大学学报(理学版), 2020, 58(6): 1429-1435.
	MU Xiangyu, FAN Yu, LI Suji, et al. Method of generating metamorphic relationship based on gradient descent algorithm[J]. Journal of Jilin University (Science Edition), 2020, 58(6): 1429-1435.
[12]	黎静华, 黄乾, 韦善阳, 等. 基于S-BGD和梯度累积策略的改进深度学习方法及其在光伏出力预测中的应用[J]. 电网技术, 2017, 41(10): 3292-3300.
	LI Jinghua, HUANG Qian, WEI Shanyang, et al. Improved deep learning algorithm based on S-BGD and gradient pile strategy and its application in PV power forecasting[J]. Power System Technology, 2017, 41(10): 3292-3300.
[13]	李兴怡, 岳洋. 梯度下降算法研究综述[J]. 软件工程, 2020, 23(2): 1-4.
	LI Xingyi, YUE Yang. Survey of gradient descent algorithm[J]. Software Engineer, 2020, 23(2): 1-4.
[14]	陈泽坤, 程晓荣. 基于梯度下降算法的房价回归分析与预测[J]. 信息技术与信息化, 2020, 5(5): 10-13.
	CHEN Zekun, CHENG Xiaorong. Housing price regression analysis and prediction based on gradient descent algorithm[J]. Information Technology & Informatization, 2020, 5(5): 10-13.
[15]	耿晓燕, 何畅, 万玉金. 基于灰色关联法的页岩气水平井产能评价及预测[J]. 数学的实践与认识, 2020, 50(19): 100-106.
	GENG Xiaoyan, HE Chang, WAN Yujin. Production evaluation and prediction of horizontal shale gas wells based on grey correlation method[J]. Mathematics in Practice and Theory, 2020, 50(19): 100-106.
[16]	郑亚军, 刘宝成, 张旭泽, 等. 数据驱动与地质规律融合的超低渗油藏产能预测方法[J]. 石油地质与工程, 2022, 36(4): 75-81.
	ZHENG Yajun, LIU Baocheng, ZHANG Xuze, et al. Productivity prediction method of ultra-low permeability reservoir based on data-driven and geological law[J]. Petroleum Geology & Engineering, 2022, 36(4): 75-81.
[17]	田媛, 梁永全. 基于小批量梯度下降的布谷鸟搜索算法[J]. 山东科技大学学报(自然科学版), 2020, 39(5): 56-67.
	TIAN Yuan, LIANG Yongquan. Cuckoo search algorithm based on mini-batch gradient descent[J]. Journal of Shandong University of Science and Technology(Natural Science), 2020, 39 (5): 56-67.
[18]	宋杰, 朱勇, 许冰. 批量减数更新方差缩减梯度下降算法BSUG[J]. 计算机工程与应用, 2020, 56(22): 117-123. doi: 10.3778/j.issn.1002-8331.2003-0261
	SONG Jie, ZHU Yong, XU Bing. Batch subtraction update variance reduction gradient descent algorithm BSUG[J]. Computer Engineering and Applications, 2020, 56(22): 117-123. doi: 10.3778/j.issn.1002-8331.2003-0261
[19]	李英, 贺春林. 面向深度神经网络训练的数据差分隐私保护随机梯度下降算法[J]. 计算机应用与软件, 2020, 37(4): 252-259.
	LI Ying, HE Chunlin. Data differential privacy protection stochastic gradient descent algorithm for deep neural network training[J]. Computer Applications and Software, 2020, 37(4): 252-259.
[20]	王一鸣, 宋先海, 张学强. 应用人工神经网络算法的地震面波非线性反演[J]. 石油地球物理勘探, 2021, 56(5): 979-991.
	WANG Yiming, SONG Xianhai, ZHANG Xueqiang. Research on nonlinear inversion of seismic surface waves based on artificial neural network algorithm[J]. Oil Geophysical Prospecting, 2021, 56(5): 979-991.
[21]	许泽坤, 陈隽. 非线性结构地震响应的神经网络算法[J]. 工程力学, 2021, 38(9): 133-145.
	XU Zekun, CHEN Jun. Neural network algorithm for nonlinear structural seismic response[J]. Engineering Mechanics, 2021, 38(9): 133-145.
[22]	卫浪, 蒲红宇, 向辉, 等. 基于改进神经网络的丙烷回收流程多目标优化[J]. 石油与天然气化工, 2021, 50(1): 66-71.
	WEI Lang, PU Hongyu, XIANG Hui, et al. Multi-objective optimization of propane recovery process based on improved BP neural network[J]. Chemical Engineering of Oil & Gas, 2021, 50(1): 66-71.

序号	参数名称	参数符号
1	水平井段长/m	SP
2	完井井段长/m	WJ
3	总液量/m³	ZY
4	总砂量/m³	ZS
5	压裂段数	YL
6	含气量/（m³/t）	HT
7	孔隙度/%	KX
8	渗透率/10^-3 μm²	ST

序号	参数名称	构建形式
1	每米液量/m³	ZY/（WJ×0.3）
2	每米砂量/m³	ZS/（WJ×0.3）
3	视砂比/m³	ZS/ZY
4	压裂单段长/m	（WJ/YL）×0.3
5	随机组合	（ZY-ZS）/YL

特征参数类型	修正决定系数（T）	均方误差（MSE）
构建特征参数前	0.68	0.536
构建特征参数后	0.95	0.025

序号	实际日产量/ 10³ m³	传统BP 方法预测值/ 10³ m³	相对误差/ %	改进的网络模型方法预测值/ 10³ m³	相对误差/ %
1	9.184	7.632	-16.89	8.863	-3.49
2	2.651	3.594	35.57	2.889	8.97
3	3.706	4.356	17.53	3.836	3.51
4	5.907	4.986	-15.59	6.106	3.36
5	5.678	6.723	18.40	5.852	3.06
6	3.188	3.976	24.72	2.901	-9.00
7	5.729	6.932	20.99	5.665	-1.12
8	9.740	8.134	-16.49	9.985	2.52
9	2.403	1.688	-29.75	2.565	6.74
10	5.374	6.266	16.59	5.594	4.09

模型类型	均方误差（MSE）	修正决定系数（T）
传统BP网络模型	0.132	0.73
改进的网络模型	0.006	0.95