人工神经网络预测木头油田储层孔隙度渗透率

Predicting Reservoir Porosity and Permeability of Mutou Oilfield with Artificial Neural Network

由于历史原因,木头小规模油田在开发过程中,评价油藏、储层的物性参数严重匮乏。随着开发阶段的不断变化,剩余油分布、注采关系分析和油藏地质建模、数值模拟等工作需要高精度的物性参数。本文提出了应用改进的人工神经网络BP模型对储层孔隙度、渗透率进行预测的方法,通过实际运用,和使用多元逐步回归法相比,预测的精度大幅度提高,渗透率相关系数可由0.8436提高到0.9961,相对误差2.19%,从而为深化储层认识提供了准确的孔隙度、渗透率参数。

Reservoir evaluation are seriously short of physical parameters during development of the small Mutou oilfield due to some historical factors.Physical parameters with high accuracy are needed for analyzing residual oil distri-bution,constructng reservoir geologic models and numeri-cal modeling along with the advancing of development.A method is presented in this paper for predicting reservoir porosity and permeability with artificial neural network BP model.Compared with the multiple regression analysis,the accuracy of prediction of this method is improved by a big margin with correlation coefficient going up from0.7886to0.9943and relative error lowering to2.19%.