Neural network approximation of resistivity tool responses in three–layer formation model

Abstract

We apply neural networks function approximation method to the problem of resistivity data modeling on the example of a three–layer geoelectric formation model. The model parameters distribution of the training database depends on the relative position of the tool coils and the layer boundaries. It is obtained on a test database that the signals calculated using neural networks coincide with the synthetic ones within one measurement error for more than 99.9% of the test samples.