Application of kinetic models and neural networks to predict the embedding rate during storage of fingered citron essential oil microcapsules

Abstract

The aim of this work was to develop kinetic models of the embedding rate during storage of fingered citron essential oil microcapsules and evaluate the performance of artificial neural network to predict the kinetic parameters. Release kinetics of microencapsulated (by β-cyclodextrin) fingered citron essential oil was investigated under a series of temperature levels (10, 25, 40, 55, 70, 85, and 100 °C). In addition, the model was trained using a back-propagation algorithm and one hidden layer artificial neural network (ANN) was employed. The results indicated that the optimal ANN model was developed when the optimal number of neurons in the hidden layer was 19. Additionally, the correlation coefficients between predicted k, t1/2, or D-value and experimental values were greater than 0.9971 in all case. Thus, the overall results showed that ANN could be a potential tool for quick and accurate prediction of the kinetic parameters. © Springer-Verlag Berlin Heidelberg 2014.