基于MATLAB支持向量机定向拟合在储层孔隙度预测中的应用

Application of Support Vector MachineDirectional Fitting Based on MATLAB in Reservoir Porosity Prediction

摘要: 针对磨溪–高石梯区块灯影组致密碳酸盐岩储层孔隙度难以预测的问题，本文在使用基于MATLAB支持向量机分类模型的储层类型预测方法的基础上，总结四种不同储集类型测井电性参数分布规律及主控电性参数，再根据分布规律划分四个板块以确定回归线走向，最后采用逐步回归法定向多元回归拟合方程分类预测储层孔隙度。以此方法对致密碳酸盐岩孔隙度进行预测，不仅预测更加精确，也能高效地完成预测工作，同时能方便快速地计算出有效储层的厚度及位置，提高产油效率。

Abstract: Aiming at the problem that it is difficult to predict the porosity of tight carbonate reservoir in Moxi-Gaoshiti area, based on the reservoir type prediction method based on MATLAB support vector machine classification model, this paper summarizes the distribution law of logging elec-trical parameters and major electrical parameters of four different reservoir types, and then di-vides four plates according to the distribution rules to determine the trend of regression line by using stepwise regression method; directional multiple regression fitting equation is used to classify and predict reservoir porosity. This method can not only predict the porosity of tight carbonate rocks more accurately, but also efficiently complete the prediction work. At the same time, it can calculate the effective reservoir thickness and position conveniently and quickly, and improve the oil production efficiency.