基于电性参数和GA-BP神经网络的砂岩孔渗预测

Prediction of Sandstone Porosity and Permeability Based on Electrical Parameters and GA-BP Neural Network

以地表岩性复杂的四川盆地为采样区,基于神经网络可解决电性参数与储层物性参数之间非线性映射问题的优势性能,讨论低孔低渗致密储层的砂岩孔渗预测方法。以密度、电阻率和极化率等电性参数作为网络模型的输入参数,利用遗传算法(GA)对BP神经网络(BPNN)的权值和阈值进行优化,进而建立并训练GA-BP神经网络模型。与传统的多元回归法相比,GA-BP神经网络模型的孔渗预测实验结果更优。

Taking the Sichuan Basin with complex surface lithology as the sampling area,the neural network can be used to solve the problem of nonlinear mapping between electrical parameters and reservoir physical parameters,and the prediction method of sandstone porosity and permeability in low-porosity and low-permeability tight reservoirs is discussed.The electrical parameters such as density,resistivity and polarizability were used as the input parameters of the network model,and the weights and thresholds of the BP neural network(BPNN)were optimized by genetic algorithm(GA),and then the GA-BP neural network model was established and trained.Compared with the traditional multiple regression method,the results of the porosity prediction experiment of the GA-BP neural network model are better.