论文部分内容阅读
针对支持向量回归机(SVR)预测孔隙度精度不高的问题,将线性拟合方法与SVR方法相结合,在模型建立过程中融入岩性信息,并为孔隙度预测值设定约束,建立一种新的储层孔隙度预测模型。建模过程中,首先利用密度测井对孔隙度进行线性拟合,然后在模型的输入属性中加入视岩性特征属性,以实际孔隙度与孔隙度拟合值差值作为预测值建立SVR预测模型,最后利用密度测井与孔隙度的图版关系对孔隙度预测值进行约束。预测结果表明,采用综合方法建立的孔隙度预测模型预测精度明显高于传统线性方法和SVR方法。该研究为储层孔隙度预测提供了一种新方法。
Aiming at the problem of poor prediction of porosity by support vector regression (SVR), the linear fitting method is combined with the SVR method to integrate lithology information during the model establishment and set a constraint on the prediction of porosity. New reservoir porosity prediction model. In the process of modeling, firstly, density log is used to linearly fit the porosity, and then lithological characteristics are added to the input properties of the model. The SVR prediction is established by using the difference between the actual porosity and the porosity fitting value Finally, the relationship between density log and porosity is used to constrain the prediction of porosity. The prediction results show that the prediction accuracy of the porosity prediction model established by comprehensive method is obviously higher than that of the traditional linear method and SVR method. This study provides a new method for reservoir porosity prediction.