地震属性模式聚类预测储层物性参数  

Prediction of physical reservoir parameters by seismic attributes pattern cluster

作　　者：陈波[1] 胡少华[2] 毕建军[1] 

机构地区：[1]中国矿业大学 [2]中国地质大学

出　　处：《石油地球物理勘探》2005年第2期204-208,共5页Oil Geophysical Prospecting

摘　　要：地震模式聚类预测技术是一种模式识别技术,在油气储层预测中有广泛的应用。其应用成功的关键在于:1所用地震资料的质量满足高信噪比、高分辨率及高保真度的要求;2建立准确的地质模型(综合框架) ,为神经网络的训练提供好的学习样本;3选用多层感知器神经网络。在上述第二个环节中,涉及储层层段划分上应力求按岩性特征细化,然后按测井细分层数据对地震信息进行准确标定,建立地震属性信息与地质、井信息的联系。文中列举了两个油田的应用实例,进一步证实了此方法的应用效果。Seismic pattern cluster prediction technique is a pattern recognition technique,which has wide application in the prediction of oil/gas reservoir.The key of successful application lies as follows:①the quality of seismic data used meet the needs of high S/N ratio,high resolution and hi-fi;②building up correct geological model (integrative framework) so that can provide good study sample for neural network training;③selecting neural network with multi-layer sensor.In above-mentioned second link,in concern with formation-partitioning of reservoir it must make effort to refine according to lithology characters,then correctly labele the seismic information according to layer-refined data by logging data and build up the link between seismic attributes information and geologic and drilling information.The paper listed the applied cases of two oilfields and further proved application effects of the method.

关 键 词：模式聚类 储层物性参数 模式识别技术 地震属性信息 神经网络 多层感知器 预测技术 储层预测 高信噪比 地震资料 高保真度 高分辨率 地质模型 学习样本 岩性特征 地震信息 应用实例 油气 测井 油田 

分 类 号：P618.13[天文地球—矿床学] TP183[天文地球—地质学]