PREDICTING CETANE NUMBER OF GASOILS FROM INFRA-RED SPECTRA USING NEURAL NETWORK

摘要

A method for prediction of cetane numbers of gasoils, wherein the following steps are carried out: a) measuring the I.R. spectra of a large set of gasoils, from a wide variety of sources; b) selecting in the spectral region a range of wave numbers; and converting a number of the wavelengths in question to absorption data and using said absorption data as an input to multivariate statistical analysis or a neural network; c) analyzing the spectral data using multivariate statistical techniques or neural networks; d) determining cetane number of the gasoil by conventional measurement; e) selecting a training data set comprising measured I.R. data and conventionally measured cetane number data and correlating the obtained absorbance values with cetane number, generating a set of predictive data; and subsequently f) applying these data to infra-red spectra, taken under the same conditions, for gasoils of unknown cetane number, thus providing the cetane number of the unknown gasoil.