Modeling of rehydration behavior of freeze- and vacuum-dried damson plums by an enhanced Chebyshev network.

The aim of this paper is to investigate the rehydration properties of freeze- and vacuum-dried damson plums (Prunus insititia) at three different temperatures (25, 45 and 60°C). First, kinetic models (Weibull, Peleg, Exponential and First-order) were designed to construct mathematical models and analyze the rehydration kinetics. Second, an artificial Chebsyhev network was designed for modeling of the rehydration kinetics such that a novel extreme learning machine-based feature extraction layer is proposed to improve its modeling capability. The experimental data and artificial models were analyzed considering the randomly selected data sets, and the root mean squared errors (RMSE) were computed to compare accuracy of the models. Due to orthogonality and feature extraction, the proposed enhanced Chebyshev network was obtained as the best approximator model among tested models with the the lowest RMSE values to explain the rehydration behavior of damson plums. While the percentage RMSE values for kinetic models vary in the range of ~3.1 and 4.8%, the maximum and minimum percentage values for Chebyshev networks are 2.32% and 0.51%, respectively. It is concluded that the proposed Chebyshev network can be used as a parsimonious model in the embedded design of the rehydration and drying machines so that predefined rehydration and drying characteristics can be accurately defined. [ABSTRACT FROM AUTHOR]