Multi-granularity temporal structure representation based outlier detection method for prediction of oil reservoir

doi:10.11772/j.issn.1001-9081.2020101867

Abstract

Abstract: The traditional methods for prediction of oil reservoir utilize the seismic attributes generated when seismic waves passing through the stratum and geologic drilling data to make a comprehensive judgment in combination with the traditional geophysical methods. However, this type of prediction methods has high cost of research and judgement and its accuracy strongly depends on the prior knowledge of the experts. To address the above issues, based on the seismic data of the Subei Basin of Jiangsu Oilfield, and considering the sparseness and randomness of oil-labeling samples, a multi-granularity temporal structure representation based outlier detection algorithm was proposed to perform the prediction by using the post-stack seismic trace data. Firstly, the multi-granularity temporal structures for the single seismic trace data was extracted, and the independent feature representations were formed. Secondly, based on extracting multiple granularity temporal structure representations, feature fusion was carried out to form the fusion representation of seismic trace data. Finally, a cost-sensitive method was utilized for the joint training and judgement to the fused features, so as to obtain the results of oil reservoir prediction for these seismic data. Experiments and simulations of the proposed algorithm were performed on an actual seismic data of Jiangsu Oilfield. Experimental results show that the proposed algorithm is improved by 10% on Area Under Curve (AUC) compared to both of the Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) algorithms.

Key words: prediction of oil reservoir, outlier detection, multi-granularity, cost-sensitive, temporal structure

摘要： 传统储层含油性勘测方法利用地震波穿过地层时产生的相关地震属性和地质钻井资料结合传统地球物理方法进行综合研判，但该类勘测方法往往存在研判成本高且对专家先验知识依赖性强的问题。针对该问题，以江苏油田苏北盆地的地震资料为基础，并结合含油样本的稀疏性和随机性，提出了一种基于多粒度时序结构表示的异常检测算法，直接利用叠后地震道数据进行预测。该算法首先对于单个地震道数据提取多粒度时序结构并形成独立特征表示；其次，在提取多个粒度时序结构表示的基础上进行特征融合，以形成对地震道数据的融合表示；最后，通过对融合后的特征采用代价敏感方法进行联合训练和判别，从而得到对于该地震数据的含油性勘测结果。所提算法在江苏油田实际原始地震资料上进行了实验仿真，实验结果表明：所提算法相比长短期记忆（LSTM）和门控循环单元（GRU）算法在曲线下方的面积（AUC）指标上均提升了10%。

关键词: 含油性检测, 异常检测, 多粒度, 代价敏感, 时序结构

CLC Number:

TP391

MENG Fan, CHEN Guang, WANG Yong, GAO Yang, GAO Dequn, JIA Wenlong. Multi-granularity temporal structure representation based outlier detection method for prediction of oil reservoir[J]. Journal of Computer Applications, 2021, 41(8): 2453-2459.

孟凡, 陈广, 王勇, 高阳, 高德群, 贾文龙. 基于多粒度时序结构表示的异常检测算法在储层含油性检测中应用[J]. 计算机应用, 2021, 41(8): 2453-2459.

References

[1] GHOLAMI R,MORADZADEH A,RASOULI V,et al. Shear wave velocity prediction using seismic attributes and well log data[J]. Acta Geophysica,2014,62(4):818-848.
[2] GRAVES A, JAITLY N, MOHAMED A R. Hybrid speech recognition with deep bidirectional LSTM[C]//Proceedings of the 2013 IEEE Workshop on Automatic Speech Recognition and Understanding. Piscataway:IEEE,2013:273-278.
[3] HINTON G E,OSINDERO S,TEH Y W. A fast learning algorithm for deep belief nets[J]. Neural Computation,2006,18(7):1527-1554.
[4] SARIKAYA R,HINTON G E,DEORAS A. Application of deep belief networks for natural language understanding[J]. IEEE/ACM Transactions on Audio,Speech,and Language Processing,2014, 22(4):778-784.
[5] LeCUN Y,BENGIO Y,HINTON G. Deep learning[J]. Nature, 2015,521(7553):436-444.
[6] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM,2017,60(6):84-90.
[7] KIROS J R, CHAN W, HINTON G E. Illustrative language understanding:large-scale visual grounding with image search[C]//Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA:Association for Computational Linguistics,2018:922-933.
[8] MENG F D,CHEN S Y,ZHANG Y C,et al. Characterization of motor oil by laser-induced fluorescence[J]. Analytical Letters, 2015,48(13):2090-2095.
[9] CUDJOE S,VINASSA M,HENRIQUE BESSA GOMES J,et al. A comprehensive approach to sweet-spot mapping for hydraulic fracturing and CO2 huff-n-puff injection in Chattanooga shale formation[J]. Journal of Natural Gas Science and Engineering, 2016,33:1201-1218.
[10] EBRAHIMI A,KHAMEHCHI E. Developing a novel workflow for natural gas lift optimization using advanced support vector machine[J]. Journal of Natural Gas Science and Engineering,2016,28:626-638.
[11] ZHANG J G. Application of gray Elman neural network to predict the gas emission amount[J]. Advanced Materials Research, 2013,706/707/708:1750-1754.
[12] 罗浩然, 尹成, 丁峰, 等. 概率神经网络的平滑参数分析及在地震属性分析中的应用[J]. 石油物探,2017,56(4):551-558. (LUO H R,YIN C,DING F,et al. The smoothing parameter analysis of probabilistic neural network and its application in seismic attribute analysis[J]. Geophysical Prospecting for Petroleum,2017,56(4):551-558.)
[13] MASOUDI P,TOKHMECHI B,JAFARI M A,et al. Application of fuzzy classifier fusion in determining productive zones in oil wells[J]. Energy Exploration and Exploitation, 2012, 30(3):403-415.
[14] 王振洲, 张春雷, 高世臣. 利用决策树方法识别复杂碳酸盐岩岩性——以苏里格气田苏东41-33区块为例[J]. 油气地质与采收率,2017,24(6):25-33.(WANG Z Z,ZHANG C L,GAO S C. Lithology identification of complex carbonate rocks based on decision tree method:an example from Block Sudong41-33 in Sulige Gas Field[J]. Petroleum Geology and Recovery Efficiency,2017,24(6):25-33.)
[15] LI Q,ZHONG H Q,WANG Y,et al. Integrated development optimization model and its solving method of multiple gas fields[J]. Petroleum Exploration and Development,2016,43(2):293-300.
[16] 印兴耀, 叶端南, 张广智. 基于核空间的模糊聚类方法在储层预测中的应用[J]. 中国石油大学学报(自然科学版),2012,36(1):53-59.(YIN X Y,YE D N,ZHANG G Z. Application of kernel fuzzy C-means method to reservoir prediction[J]. Journal of China University of Petroleum (Edition of Natural Science), 2012,36(1):53-59.)
[17] 刘瑞林, 马在田. 神经网络在油气评价和预测方面的研究现状[J]. 地球物理学进展,1995,10(2):75-84.(LIU R L,MA Z T. The current studies of application of neural networks to oil and gas evaluation and prediction[J]. Progress in Geophysics,1995, 10(2):75-84.)
[18] 陆文凯, 牟永光. 利用BP神经网络进行测井资料外推[J]. 石油地球物理勘探,1996,31(5):712-715.(LU W K,MOU Y G. Logging data extrapolation using BP neural network[J]. Oil Geophysical Prospecting,1996,31(5):712-715.)
[19] 张, 郑晓东, 李劲松, 等. 基于SOM和PSO的非监督地震相分析技术[J]. 地球物理学报,2015,58(9):3412-3423.(ZHANG Y,ZHENG X D,LI J S,et al. Unsupervised seismic facies analysis technology based on SOM and PSO[J]. Chinese Journal of Geophysics,2015,58(9):3412-3423.)
[20] 张向君, 李幼铭, 刘洪. 神经网络结构风险最小油气预测[J]. 石油地球物理勘探,2002,37(1):73-76.(ZHANG X J,LI Y M, LIU H. Oil and gas prediction by neural networks with structure-risk-minimum[J]. Oil Geophysical Prospecting,2002, 37(1):73-76.)
[21] MOHEBBI A,KAMALPOUR R,KEYVANLOO K,et al. The prediction of permeability from well logging data based on reservoir zoning, using artificial neural networks in one of an Iranian heterogeneous oil reservoir[J]. Petroleum Science and Technology,2012,30(19):1998-2007.
[22] 张长开, 姜秀娣, 朱振宇, 等. 基于支持向量机的属性优选和储层预测[J]. 石油地球物理勘探,2012,47(2):282-285. (ZHANG C K,JIANG X D,ZHU Z Y,et al. Attributes selection and reservoir prediction based on support vector machine[J]. Oil Geophysical Prospecting,2012,47(2):282-285.)
[23] 王波, 夏同星, 谭辉煌. 基于斑块饱和模型井控属性融合法油气检测[J]. 石油物探,2017,56(2):288-294.(WANG B,XIA T X,TAN H H. The well-controlled attributes fusion method for hydrocarbon detection based on patchy-saturation model[J]. Geophysical Prospecting for Petroleum,2017,56(2):288-294.)
[24] 袁照威, 陈龙, 高世臣, 等. 基于马尔可夫-贝叶斯模拟算法的多地震属性沉积相建模方法——以苏里格气田苏区块为例[J]. 油气地质与采收率,2017,24(3):37-43.(YUAN Z W,CHEN L,GAO S C,et al. A method of sedimentary facies modeling through integration of multi-seismic attributes based on MarkovBayes model:an example from Su10 area in the north of Sulige gas field[J]. Petroleum Geology and Recovery Efficiency,2017,24(3):37-43.)
[25] 宋建国, 高强山, 李哲. 随机森林回归在地震储层预测中的应用[J]. 石油地球物理勘探,2016,51(6):1202-1211.(SONG J G,GAO Q S,LI Z. Application of random forests for regression to seismic reservoir prediction[J]. Oil Geophysical Prospecting, 2016,51(6):1202-1211.)
[26] SAIKIA P,BARUAH R D,SINGH S K,et al. Artificial neural networks in the domain of reservoir characterization:a review from shallow to deep models[J]. Computers and Geosciences,2020, 135:No. 104357.
[27] 林年添, 付超, 张栋, 等. 无监督与监督学习下的含油气储层预测[J]. 石油物探,2018,57(4):601-610.(LIN N T,FU C, ZHANG D,et al. Supervised learning and unsupervised learning for hydrocarbon prediction using multiwave seismic data[J]. Geophysical Prospecting for Petroleum,2018,57(4):601-610.)
[28] 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 N T,ZHANG D,ZHANG K,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.)
[29] 付超, 林年添, 张栋, 等. 多波地震深度学习的油气储层分布预测案例[J]. 地球物理学报,2018,61(1):293-303.(FU C, LIN N T,ZHANG D,et al. Prediction of reservoirs using multicomponent seismic data and the deep learning method[J]. Chinese Journal of Geophysics,2018,61(1):293-303.)
[30] 安鹏, 曹丹平, 赵宝银, 等. 基于LSTM循环神经网络的储层物性参数预测方法研究[J]. 地球物理学进展,2019,34(5):1849-1858.(AN P,CAO D P,ZHAO B Y,et al. Reservoir physical parameters prediction based on LSTM recurrent neural network[J]. Progress in Geophysics,2019,34(5):1849-1858.)