Geographical origin classification of petroleum crudes from near-infrared spectra of bitumens

Petroleum crudes of different geographical origin exhibit differences in ch emical composition that arise from formation and ripening processes in the crude. Such differences are transmitted to the fractions obtained in the pr ocessing of petroleum. The use of unsupervised classification/sorting metho ds such as principal component analysis (PCA) or cluster analysis to near-i nfrared (NIR) spectra for bitumens obtained from petroleum crudes of divers e origin has revealed that composition differences among bitumens are clear ly reflected in the spectra, which allows them to be distinguished in terms of origin. Accordingly, in this work we developed classification methods b ased on soft independent modeling of class analogy (SIMCA) and artificial n eural networks (ANNs). While the latter were found to accurately predict th e origin of the crudes, SIMCA methodology failed in this respect.