基于独立分量分析的地震属性优化

随着储层预测要求精度的提高，从众多地震属性集中挑选对所预测对象最敏感的地震属性，进行地震属性优化的工作相当重要。K-L变换通过一正交变换优选出一些不相关的地震属性，而不能优选出一些更高阶的相互独立的地震属性，即K-L变换仅利用了属性的二阶统计特性。独立分量分析（ICA）作为分解观测数据中独立信息的有力工具，不仅利用了信号的二阶统计特性，而且还利用了信号的高阶统计特性。将ICA引入到地震多属性优化中，利用它对地震属性进行高阶统计特征分析，从而能优选出最敏感的、相互独立的地震属性。另外，ICA属性优化方法不需要测井数据、井旁储层段的参数、钻井数据等，也不受勘探、开发阶段测井资料的数量限制。实际资料的应用分析表明，应用独立分量分析优化后的地震属性作储层预测具有较高的精度和可靠性，是一种新的属性优化方法。

A COMBINATIONAL OPTIMUM METHOD OF SEISMIC ATTRIBUTES BASED ON INDEPENDENT COMPONENT ANALYSIS

Along with the high accuracy required by the reservoir prediction, choosing the most sensitive (in other words， the most effective or the most presentative) to the targets predicted from numerous seismic attributes, carrying on the seismic attributes combination optimization is inevitably essential works to do. The K L transformation can only optimize some irrelated seismic attributes through an orthogonal transformation, but cannot optimize some higher order and relatively independent seismic attributes because the K L transformation has only used the 1st and 2nd order statistics of seismic attributes. The independent component analysis (ICA) ,used as a powerful tool separating independent information from observation data, has used not only the 1st and 2nd order, but also higher order statistics of the seismic attributes, and can optimize higher order and relatively independent seismic attributes. The seismic attributes combination optimization based on ICA does not need the well logging data, the nearby well reservoir parameter, the well drilling data etc., and is not restricted by the number of well logging data in phase of exploitation. The applied analysis of actual dada shows that the seismic attributes which are optimized based on ICA can make reservoir prediction much more reliable and with higher precision, and it is a kind of brandnew attribute combination optimization method.