Modeling some drying characteristics of sour cherry (Prunus cerasus L.) under infrared radiation using mathematical models and artificial neural networks

Abstract

The effect of air temperature, air velocity and infrared (IR) radiation on the drying kinetics of sour cherry was investigated using a laboratory infrared dryer.  Experiments were conducted at air temperatures of 35, 50 and 65°C, air velocities of 0.5, 1.1 and 1.7 m/s and IR radiations of 500, 1,000 and 1,500 W.  Five empirical drying models for describing time dependence of the moisture ratio change were fitted to experimental data.  Artificial neural network (ANN) method was used to predict the effective moisture diffusivity and specific energy consumption of the samples.  Among the applied models, Midilli et al. model was the best to predict the thin layer drying behavior of sour cherry.  Effective moisture diffusivity of sour cherry varied between 1.17×10^-10 and 8.13×10^-10 m²/s.  Activation energy of sour cherry was in the range of 30.31–     41.68 kJ/mol.  Specific energy consumption was in the range of 56.12–891.16 MJ/kg.  After well training of the ANN models, it proved that the ANN model was relatively better than the empirical models.  The best neural network feed and cascade forward back-propagation topologies for the prediction of effective moisture diffusivity and energy consumption were the 3-2-3-1 and 3-3-3-1 structures with the training algorithm of trainlm and threshold functions of tansig, tansig-logsig-tansig, respectively.  The best R² value for predication of moisture diffusivity and energy consumption were 0.9944 and 0.9905, respectively. 

Keywords: sour cherry, drying, effective moisture diffusivity, activation energy, artificial neural network