利用径向基函数预测岩石渗透率及流体分子组分

测井反问题研究中,经常遇到正演模型未知、经验公式的相关系数较小的情形,且现有的反问题处理方法难以获得理想的效果。径向基函数插值方法是目前最有效的反问题处理方法之一,该技术能够生成平滑连续的多个变量的多元函数,逼近未知的正演模型。介绍径向基函数插值方法原理,针对核磁共振实验的T2分布,采用径向基函数插值方法预测了岩石渗透率及流体分子组分,实现了多维到一维映射及多维到多维映射的2种情况,并且预测结果具有较高精度,测量值的对比验证了该方法解决测井反问题的有效性。

Prediction of Rock Permeability and Fluid Molecular Composition Using Radial Basis Function

In logging inverse problem research,the forward models are often unknown and correlation coefficients of empirical equations are small,and the existing inverse problem approaches are difficult to obtain the desired results.Radial basis function(RBF) interpolation is one of the most effective approaches to deal with inverse problems,which can generate smooth and continuous multivariate functions of many variables to approximate the forward mode.Introduced is the theory of RBF interpolation.According to T2 distributions of the NMR experiments,predicted are rock permeability and fluid molecular compositions using RBF interpolation method.Achieved are mapping from multi-dimensional to one-dimensional and from multi-dimensional to multi-dimensional,and prediction accuracy is high.Verified is the validity of RBF interpolation method to solve the logging inverse problems by comparing the predicted results with measurements.