Kinetics and Artificial Neural Network Prediction of Pistachio Drying in an Infrared Assisted Solar Dryer

Abstract

To explain the drying behavior of pistachio nuts in an infrared assisted solar dryer, nine mathematical models were fitted to the experimental data and their comparison criteria were calculated. Meanwhile, different Artificial Neural Networks (ANNs) were tested to obtain the best network for final moisture content prediction. The results showed that drying time shortened by 30% when the air temperature raised from 45 to 65oC. Increasing the IR power to 500 W caused a 45% reduction in drying time. The Henderson and Pabis model was selected as the best mathematical model to describe the drying behavior of pistachio nuts. Among the networks, Levenberg–Marquardt back-propagation training algorithm presented the best fit to the drying data with RMSE of 0.0035 and R2 of 0.999 for the training and RMSE of 0.0038 and R2 of 0.996 for the testing. Based on RMSE criteria, the ANN modeling yielded a more accurate prediction compared to all of the empirical equations.