Prediction of the Derived Cetane Number and Carbon/Hydrogen Ratio from Infrared Spectroscopic Data

A model for the prediction of derived cetane number (DCN) and carbon/hydrogen ratio (C/H) hydrocarbon mixtures, diesel fuels, diesel–gasoline blends has been developed on basis infrared (IR) spectroscopy data pure components. IR spectra 65 neat species were used to generate 127 by averaging their components a molar basis. The 44 real fuels calculated using n-paraffin, isoparaffin, olefin, naphthene, aromatic, oxygenate (PIONA-O) class averages It is shown that this strategy retains knowledge C/H, an important indicator chemical structure. Three methods compared assess DCN C/H from assembled spectra, i.e., partial least squares regression (PLSR), support vector machine (SVM), artificial neural network (ANN). was found ANNs gave best performance with errors ±1.1 average ?0.8%. Lasso-regularized linear models also find simple combinations wavenumbers yield acceptable estimations.