基于PNN的多地震属性砂体含气性预测方法及应用  

Gas Prediction Method and Application of Multi-Seismic Attributes Based on PNN

作　　者：郭振华[1] 周兆华[1] 张满郎 郑国强[1] 谷江锐[1] 

机构地区：[1]中国石油勘探开发研究院廊坊分院,河北廊坊065007

出　　处：《工程地球物理学报》2011年第5期594-599,共6页Chinese Journal of Engineering Geophysics

摘　　要：苏里格气田东区盒8段和山西组砂体含气后与泥岩阻抗值差别较小,直接用波阻抗反演预测含气性可靠性较差。本文通过分析有效储层的测井响应特征,虚拟含气性指示曲线Psg,筛选地震属性并进行相关性分析,建立了一种由测井和多地震属性组成的概率神经网络方法,来对储层的含气性进行预测,避免了单属性多解性和多属性综合判别精度低的缺点。在苏里格气田东区的应用表明,在多地震属性分析的基础上,采用概率神经网络方法对砂体的含气性进行预测是一种较好的方法手段。Because of the small difference of impedance between Gas-bearing sandstone and mudstone in H8 member and Shanxi group in SuliGe gas field,the result is less reliable by using impedance inversion directly to predict gas reservoir.This paper analyzes the characteristics of effective reservoir log responses,generates a virtual gas-bearing instructions curve named Psg,filter and correlate seismic attributes,and establish a PNN method to predict gas reservoir by integrating log curves and multi seismic attributes.The calculation results show that this method is effective to avoid the multiplicity of a single seismic attribute and the difficulty in recognition of integrated multiple seismic attributes.This method is effective to be proved in Sulige gas field.

关 键 词：苏里格气田 概率神经网络 多地震属性 含气性预测 

分 类 号：P631.4[天文地球—地质矿产勘探]