The crystallization prediction of tripalmitin in triolein: an artificial neural network (ann) approach

Abstract

The use of classical theories in lipid crystallization (i.e. Avrami model), give only a partial fit of the experimental data. This fact was explained because its was shown a drastic drop in the values of the interfacial free energy of the blends, but in the tripalmitin pure not. An alternative to the modeling to this type of systems was demonstrated with a one type of ANN. It compares the predictions of the ANN Vs the predictions of the Avrami model. The predictions of the ANN were good in all cases in the levels of cooling was low, but Avrami cannot fit the experimental data. The analysis of the ANN shown the possibility of the presence of two mechanism of crystallization associated to the cooling.