Oil-gas well tragectory modeling on the base of radial basis function networks

Abstract

The problem of oil-gas well trajectory modelingis considered. The possibility of the effective drill bit motion model identification on the base of the neural network technologies is shown. The optimal approach to structural and parametrical identification of the drill bit neural network model based on use of two RBF-networks is investigated. The approach to smoothing and reconstruction of the measured data on the base of auxiliary smoothing RBF-networks regularized by the special criterion is suggested. The example of the neural network model identification is shown.