Chemometric techniques were used to extract relevant information from near infrared (NIR) spectral data to accurately classify physical properties of complex fuel samples. Discrimination of fuel types and classification of flash point, freezing point, and boiling point of jet fuels were investigated. Optimal partial least squares discriminant analysis (oPLS-DA), fuzzy rule-building expert system (FuRES), and support vector machine (SVM) were used to build the calibration models between the NIR spectra and classes of physical property of jet fuels. oPLS-DA, FuRES, and SVM were compared with respect to prediction accuracy. The results indicated that combined with chemometric classifiers NIR spectroscopy would be a fast method to monitor the changes of jet fuel physical properties.
Six widely used approaches of preprocessing NIR spectra were compared with respect to property prediction of jet fuels by NIR spectroscopy. These approaches included calculating the derivatives of spectra, multiplicative signal correction (MSC), standard normal variate (SNV) transformation, orthogonal signal correction (OSC), and two feature selection methods interval partial least squares (iPLS) and genetic algorithm (GA). Partial least squares (PLS) and temperature-constrained cascade correlation network (TCCCN) were used to build the calibration model and the prediction performance are compared. The validation of the calibration model was conducted by applying the bootstrapped Latin partition method that can give a measure of the precision.
Chemometric tools were used to determine the concentrations of the main products namely triolein and trielaidin in the mixtures of thermally treated triolein, a naturally occurring glyceride of oleic acid. The products formed during the thermal treatment at each temperature had been analyzed both by infrared spectrometry and gas chromatography/mass spectrometry (GC/MS). The GC/MS analysis was performed after derivatization of the fatty acids into their methyl esters (FAMEs). The combined analysis revealed that the thermal treatment induces not only cis–trans isomerization but also fission and fusion in the molecules.
A regularized baseline correction method that uses basis set projection to estimate spectral backgrounds had been developed and applied to GC/MS data. An orthogonal basis was constructed using singular value decomposition (SVD) for each GC/MS two-way data object from a set of baseline mass spectra. The novel component of this method was the regularization parameter that prevents overfitting that may produce negative peaks in the corrected mass spectra or ion chromatograms. The parameters for baseline correction were optimized so that the projected difference resolution (PDR) or signal-to-noise ratio (SNR) was maximized. This new baseline correction method was evaluated with two synthetic data sets and a real GC/MS data set. The prediction accuracies obtained by using the FuRES and PLS-DA as classifiers were compared. The results indicated that baseline correction of the two-way GC/MS data using the proposed methods resulted in a significant increase in average PDR values and prediction accuracies.