Drying kinetics study of parboiled rice by using artificial neural network model

The objective of this research was to predict hybrid hot air-infrared radiation drying kinetics of Leb Nok Pattani parboiled rice using a mathematical model and an artificial neural network model. Drying kinetics of parboiled rice was investigated considering different drying conditions. The drying experiments were performed at three levels of drying air temperatures of 60-100°C, two levels of infrared intensity of 5,118 and 7,678 W/m², air velocity was fixed at 1±0.2 m/s. Nine different mathematical models available in the literature were fitted to the experimental data. Among the considered mathematical drying models, Two-term model was found to be more suitable for predicting drying of parboiled rice. In addition, a feed-forward Levenberg-Maqurdt´s propagation artificial neural network was employed to determine the relationship between the moisture ratio of the material to be dried and the input parameters of the drying time, air temperature and infrared intensity. A selected neural network structure was used for studying the influence of transfer function in hidden, output layer and the number of hidden neurons. The results between mathematical model and artificial neural network model were compared with the experimental data. In this research, it was obviously found that the prediction results of feed-forward Levenberg-Maqurdt´s propagation with four nodes in one hidden layer, logarithmic sigmoid (logsig) and hyperbolic tangent sigmoid (tansig) transfer function in hidden and output layer, respectively, are good agreement with experimental results than empirical model. Thus, it was concluded that the artificial neural network could be effective for modeling of the grain drying process. For quality of parboiled rice drying, the result reveals that was insignificantly different with different drying temperature and infrared intensity. Increases in drying temperature result in whiteness and specific energy consumption decrease.