Density log construction using machine learning
Proceeding of the 17th International Congress of the Brazilian Geophysical · 2021
Determination of the reflection coefficients is a key element to a well-to-seismic tie, and the density log has a major petrophysical importance as it is used to calculate the acoustic impedance and the reflectivity log.Many authors have developed empirical relations to determine the bulk density log from other logs information such as compressional velocity and shale volume fraction.Machine learning techniques have been applied in image and voice recognition, medical diagnosis, statistics and many other problems involving regression and classification including some in geophysics.The idea in this work is to compare the accuracy of an Artificial Neural Network model, which calculates the density log by having other logs as input, with the existing empirical models, and determine which one presents the best adjustment.An ANN model was created and the comparison with the empirical models was made by statistical analysis such as calculating the mean squared error, the relative error and correlation factor.The ANN model presented smaller errors and higher precision on the adjustment compared to the empirical models.