Study on Quantitative Structure–Retention Relationships for Hydrocarbons in FCC Gasoline

Abstract

A series of hydrocarbons in FCC gasoline have been used to develop quantitative structure-retention relationships (QSRR) for their gas chromatographic retention index (RI) by molecular descriptors which were calculated by Dragon software. QSRR models were built by adopting multiple linear regression (MLR) and artificial neural network (ANN). However, the results showed more or less the same quality with the predictive correlation coefficient R of 0.9952 and 0.9953 for MLR and ANN respectively. The obtained results showed that the linear method is satisfactory to model the gas chromatographic retention index at least to the current dataset.