融合深度置信网络与与核极限学习机算法的核磁共振测井储层渗透率预测方法

发布时间：2018-02-11 15:31

本文关键词： 深度学习核磁共振测井渗透率深度置信网络深度置信-核极限学习机　出处：《计算机应用》2017年10期 　论文类型：期刊论文

【摘要】：由于低孔低渗储层孔隙结构较为复杂,现有核磁共振(NMR)测井渗透率模型对于低孔低渗储层预测精度不高。为此,提出一种融合深度置信网络(DBN)算法与核极限学习机(KELM)算法的渗透率预测方法。该方法首先对DBN模型进行预训练,然后将KELM模型作为预测器放置在训练好DBN模型后,利用训练数据进行有监督的训练,最终形成深度置信-核极限学习机(DBKELMN)模型。考虑到该模型需充分利用反映孔隙结构的横向弛豫时间谱信息,将离散化后的核磁共振测井横向弛豫时间谱作为输入,渗透率作为输出,确定NMR测井横向弛豫时间谱与渗透率的函数关系,并基于该函数关系对储层渗透率进行预测。实例应用表明,融合DBN算法与KELM算法的渗透率预测方法是有效的,预测样本的平均绝对误差(MAE)较斯伦贝谢道尔研究中心(SDR)模型降低了0.34。融合DBN算法与KELM算法的渗透率预测方法可提高低孔渗储层渗透率预测精度,可应用于油气田勘探开发。
[Abstract]:Due to the complexity of pore structure in low porosity and low permeability reservoirs, the existing nuclear magnetic resonance (NMR) logging permeability model is not accurate in predicting low porosity and low permeability reservoirs. This paper presents a permeability prediction method based on the fusion of depth confidence network (DNN) algorithm and kernel limit learning machine (KELM) algorithm. Firstly, the DBN model is pre-trained, and then the KELM model is placed as a predictor after the DBN model is trained. Using the training data for supervised training, the depth confidence nuclear limit learning machine (DBKELMN) model is formed. Considering that the model needs to make full use of the transverse relaxation time spectrum information which reflects the pore structure, The transversal relaxation time spectrum of NMR logging is taken as input and permeability as output, and the function relationship between transverse relaxation time spectrum and permeability of NMR log is determined. Based on the function relation, the reservoir permeability is forecasted. The application shows that the permeability prediction method based on DBN algorithm and KELM algorithm is effective. The mean absolute error (mae) of the predicted samples is 0.34 less than that of the Schlumbersheidel Research Center. The permeability prediction method combining the DBN algorithm and the KELM algorithm can improve the prediction accuracy of the permeability of the low porosity and permeability reservoirs and can be applied to the exploration and development of oil and gas fields.
【作者单位】：油气资源与勘探技术教育部重点实验室(长江大学);长江大学地球物理与石油资源学院;
【基金】：湖北省自然科学基金资助项目(2013CFB396) 中国石油天然气集团公司重大专项(2013E-38-09) 长江大学教育部实验室开放基金资助项目(K2016-09)~~
【分类号】：P631.81;TP181

【相似文献】