应用神经网络方法建立单井物性解释模型  

Establish Single Well Interpretation Models by the Application of Neural Network Technology

作　　者：赵伟祥[1] 付洁[2] 申梅英[1] 刘建云[3] 

机构地区：[1]中原石油勘探局测井公司 [2]中原油田分公司勘探开发科学研究院 [3]中原石油勘探局培训中心

出　　处：《断块油气田》2005年第5期47-48,共2页Fault-Block Oil & Gas Field

摘　　要：濮城油田南区沙二上2+3油藏,储层物性复杂,常规解释模型精度低,难以用于本区单井物性解释。采用神经网络(BP)方法按物性的自然分区特征建立了解释模型,进行分段解释。物性分段解释的孔隙度、渗透率误差比笼统解释的明显减小,为寻找剩余油提供有利的地质依据。The ( 2+3) reservoir of upper S2 in Pucheng oilfield has complicated properties, the conventional interpretation models of low accuracy can hardly be used to the interpretation of the properties of wells in this area. Therefore, interpretation models have been built using neural network technology according to the naturally divisional characteristics of the reservoir to interpret section by section. By preprocessing well log data and interpreting properties, error of porosities and permeabilities interpreted section by section is obviously cut down compared with that by general interpretations, which provides a favorable condition for geological modeling and seeking for the residual oil.

关 键 词：神经网络 储层物性 解释模型 孔隙度 渗透率 误差 神经网络方法 储层物性 解释模型 单井 应用 濮城油田 模型精度 分区特征 地质依据 

分 类 号：P618.130.2[天文地球—矿床学] TP183[天文地球—地质学]