Abstract
In this study, a chemometric strategy was developed for analysis of gas chromatographic (GC) and infrared spectroscopic (FT-IR) fingerprints of nine crude oil samples from the main oil wells of Iran to classify them and to find their origins. In this regard, a fractionation method based on saturated, aromatic, resin, and asphaltene (SARA) test was used. Then, these fractions were analyzed by GC-FID and GC–MS. Also, nine crude oil samples were analyzed by FT-IR.The obtained GC fingerprints were aligned using correlation optimized warping (COW) and auto-scaled, and then analyzed using principal component analysis (PCA) and hierarchical cluster analysis (HCA). Evaluation of PCA scores plot (explaining 89.35% of variance for two PCs) and HCA dendrogram showed that aromatic fractions belong to three classes. The clustering results of aliphatic and resin fractions also showed the presence of three clusters but with different samples due to the difference in their composition. The results of unsupervised classification were then used as a starting point for partial least squares-discriminant analysis (PLS-DA) and counter propagation-artificial neural network (CP-ANN).On the other side, FTIR fingerprints of crude oil samples were also clustered using PCA and HCA. Evaluation of PCA scores plot (explaining 95.15% of variance for two PCs) and HCA dendrogram showed that samples belong to three clusters. These initial results were used as a starting point for PLS-DA and CP-ANN. Accuracy of PLS-DA was 0.890 for calibration and 0.818 for test sets. Also, CP-ANN accuracy values were 1.000 and 0.818 for train and test sets, respectively.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call