Apports De L’analyse Statistique Et De L’intelligence Artificielle Pour Prédiction De La Perméabilité Des Réservoirs Hétérogènes : Cas Du Réservoir Cambro-ordovicien (zone 17), Hassi Messaoud, Algérie.

Résumé: In this study, the zone 17 of the Hassi Messaoud oil field (southern Algeria) was the subject of this statistical and artificial intelligence study in order to characterize the reservoir and predict the permeability in the drains on the basis of the logging data. The statistical characterization was carried out on 1668 samples, representing 6 drill holes. Data analysis was performed using monovariate, bivariate, multivariate, Principal Component Analysis (PCA) and Factorial Discriminate Analysis (FDA). The obtained results showed a heterogeneous distribution of all the petrophysical parameters such as the permeability in the drains of this reservoir. The application of artificial intelligence techniques (the genetic algorithm combined with multiple linear regression, SVM support vector machine and ANN the neuron networks) has been done on the D5 drain separately as an example showing performance R2 =0.8573. Machine Learning techniques present a very powerful tool for saving time and money, because can solve complex problems such as prediction of permeability in compacted and heterogeneous reservoirs.

Publié dans la revue: