A hybrid MPSO-BP-RBFN model for reservoir lateral prediction

Abstract

The degree of success of many oil and gas drilling, completion, and production activities depends on the accuracy of the models used in the reservoir lateral prediction and description. In this paper, a hybrid MPSO-BP-RBFN model for predicting reservoir from seismic attributes is proposed. The model in which every particle consists of binary and real parts is able to simultaneously search for optimal network topology (the number of hidden nodes) and parameters, as it proceeds. The model has been used to reservoir lateral prediction of a reservoir zone and proved the model's applicability. © 2009 Springer Berlin Heidelberg.