RY凹陷致密砂岩神经网络产能分类预测方法  

Neural Network Productivity Prediction in Tight Sandstone of RY Depression

作　　者：李玮[1] 张占松[1] 祗淑华 丁一[1] 安继星 王晓林 

机构地区：[1]长江大学地球物理与石油资源学院,湖北荆州434023 [2]中国石油集团测井有限公司华北事业部,河北任丘062552

出　　处：《测井技术》2013年第6期688-693,共6页Well Logging Technology

基　　金："十二五"中国石油天然气集团公司重大科技项目:碎屑岩储层测井产能预测技术研究及软件开发(2011B-4007)

摘　　要：研究了RY凹陷的储层沉积特征、物性特征和油水层测井响应特征,分析认为其古近系沉积相类型丰富多样,属中低孔隙度中低渗透率砂岩,储层物性复杂且变化大,平面径向流产能评价方法效果有限。从岩石物性出发,提出神经网络分类评价方法。利用排驱压力划分3类储层,以降低物性变化对产能带来的影响;再利用神经网络非线性拟合的独特优势,分别建立3类物性储层测井参数和产能关系。经新探井试油数据检验,该方法的产能预测符合率大于87.5%,可满足生产需求。Abstract： This paper studies tne response characteristics of RY depression. On account of the complex depositional facies of low porosity and permeability sandstone in RY depression and the complication of reservoir physical properties, the plane radial flow formula is not suitable in this area. The paper proposes a classified neural network method. First, the reservoirs are divided into three types according to the expulsion pressure of the reservoir to reduce the effect of complex reservoir physical properties on production prediction. Then, the productivity prediction models for reservoirs in each type are proposed with neural network method. Compared with well testing result, the method has a coincidence rate of more than 87.5 %and meets the production requirement.

关 键 词：测井解释 产能预测 神经网络 致密砂岩 孔隙结构 毛细管压力曲线 

分 类 号：P631.84[天文地球—地质矿产勘探]