Petroleum reservoir properties estimation using neural networks

In this study, seismic attributes have been used to estimate well logs in one of the Iranian petroleum reservoirs. Three static methods have been evaluated: the linear model, the multilayer perceptron (MLP) and the radial basis function (RBF). For linear case, the selection of appropriate attributes was determined by forward selection and for nonlinear one, the selection was based on the genetic algorithm (GA) result. Parameters of nonlinear models were determined by cross-validation and then well logs were estimated. By comparing estimated and actual logs, RBF has the best performance with least training error. Since well logs contain high frequency content, so localized networks such as RBF has better performance than MLP through the study data set.