APLICACIÓN NO CONVENCIONAL DE REDES NEURONALES PARA PREDECIR PROPIEDADES PETROPHYSICA EN UN CUBO SISMICO NON CLASSICAL USE OF NEURONAL NETWORKS TO PREDICT PETROPHYSICA PROPIERTIES IN A SEISMIC CUBE

The methodologies used in the industry to predict petrophysical properties through the seismic data and well data are based on neural network algorithms or trace inversion process. The prediction process is based on training of a neural network. The input are seismic attributes, petrophysical property, and lithologcal data extracted from the desired output well. This paper, proposes use as input of the neural network (NN) as the set of seismic attributes calculated from a previously interpreted horizon, i.e. instead of to work with the seismic cube, work with seismic attributes extracted from the seismic, horizon, in the zone of interest. The results, in this case, are better than those produced by conventional approaches.