Quantification models based on the processing of FTIR spectra by partial least squares regression (PLS) were created in order to develop a method for the determination of 2-ethylhexyl nitrate (2-EHN) in diesel fuels. The set of standards was prepared using 2-EHN, biodiesel (FAME) and various mineral diesel fuels (2-EHN free). The standards were prepared in the concentration range of 2-EHN of 0-2436mgkg-1. The set of the standards was divided into the calibration, validation and test sets. While the calibration set was used to build the model, validation set was used in order to optimize the model parameters. The test set of the standards was used to assess the predictive ability and repeatability of the model. Several hundreds of various models were developed and compared in order to find a suitable combination of the preprocessing methods and number of latent variables. The most promising model was developed using mean centered spectra in the form of their first derivative and smoothed using Gap-Segment derivative. The model showed quite good predictive ability and repeatability.

University of Chemistry and Technology Prague Department of Petroleum Technology and Alternative Fuels Technicka 5 166 28 Prague 6 Czech Republic

Citace poskytuje Crossref.org