ANN-based models for moisture diffusivity coefficient and moisture loss at equilibrium in osmotic dehydration process

Equations were developed using artificial neural networks to predict water diffusivity coefficient (D-e) and moisture loss at equilibrium point (ML infinity) in order to get the moisture loss (ML) at any time in osmotic dehydration of fruits. These models mathematically correlate nine processing variables (temperature and concentration of osmotic solution, water and solid composition of the fruit, porosity, surface area, characteristic length, solution-to-fruit mass ratio, and agitation level) with D-e and ML infinity. Models were developed using a wide variety of data from the literature and they were able to predict D-e (r = 0.98) and ML infinity (r = 0.94) in a wide range of variable conditions. With these two parameters known, ML can be calculated using Crank's solutions of Fick's second law. The wide range of processing variables considered for the formulation of these models, and their easy implementation in a spread-sheet, using a set of equations, makes them very useful and practical for process design and control.