随机森林在储层孔隙度预测中的应用

Application of Random Forest in reservoir porosity prediction

传统的储层孔隙度计算主要采用统计回归的方法,但是在实际环境中,储层状况复杂,非均质性较强,采用传统方法所计算出的储层孔隙度误差较大。针对以上问题,本文在基于测井数据的基础上,将随机森林方法引入到储层孔隙度预测中,建立测井数据与储层孔隙度之间的非线性关系,实验证明该方法预测的储层孔隙度误差较小。与多元线性回归相比,能有效提高储层测井解释模型的精度,为储层综合评价提供可靠的地质参数。

Traditional reservoir porosity calculation mainly adopts statistical regression method. However,in the real environment,reservoir conditions are complex and heterogeneity is strong. The reservoir porosity error calculated by the traditional method is large.In viewof the above problems,based on the logging curve,a Random Forest method is introduced into the prediction of reservoir porosity to establish a nonlinear relationship between logging data and reservoir porosity. The experimental results showthat the reservoir porosity error predicted by this method is less. Compared with multiple linear regression,it can effectively improve the accuracy of reservoir logging interpretation model and provide reliable geological parameters for reservoir comprehensive evaluation.