粗集理论在地震储层预测中的应用

Application of the rough set theory to seismic reservoir predication

在地震储层预测中,可采用的地震属性种类繁多,但太多地震属性常常会起到干扰作用,影响储层的预测精度.因此,为提高地震储层预测精度,把粗糙集理论融入到地震属性的优化中,利用粗糙集理论所具有的提取有用属性、简化信息处理的能力,优选出地震属性中的敏感属性是本文的研究目的.本文采用了一种基于属性方差的自组织神经网络量化方法,并运用基于区别矩阵的属性频率约简算法对地震属性进行优选.实例分析表明：该方法可行有效,可以最大限度地删除冗余地震属性,用优选出的敏感属性组合对多种储层参数进行预测均已取得了较好的效果.

In the study of seismic reservoir predication,lots of seismic attributes are utilizable,but too many attributes could affect the accuracy of reservoir predication.So,for the sake of improving the accuracy of seismic reservoir predication,this stady utilizes the rough set theory to extract useful attributes and simplify information processing,to sort out significative attributes from seismic attributes.We use a self-organized neural network quantization method that is dependent on the variance of seismic attributes,and optimize seismic attributes by the discernibility matrix attributes frequency reduction algorithm.The example indicates：that this method is very effective,and can delete utmostly redundant seismic attributes.The multiple reservoir parameters are well predicated by significative attributes is very good.