Simulation study of oil and water migration modeling based on radial basic probabilistic neural network

Abstract

An actual physical simulation model is constructed to simulate the course of oil and water migration. Under certain physical property conditions, we simulate the water injection well and the oil well on the physical simulation model, and continuous measured online the oil and water content of different area of model in three-dimensional space using the 512 routes resistivity measuring circuit, then we can obtain large numbers of simulation samples. Considering the issues that the relationship between the remaining oil and every parameters of water displacing oil is complicated and nonlinear and the radial basic probabilistic neural network has the ability of strong nonlinear function approach and fast convergence, in this paper, the radial basic probabilistic neural network is used to establish the oil and water migration model. We construct the structure of radial basic probabilistic neural network, and adopt the K-Nearest Neighbor algorithm and least square method to train the network. The experimental results show that this method is feasible and effective.