Modeling and parameter optimization of rough rice drying using artificial neural networks

An artificial neural network was developed for modeling of rough rice drying. The ANN outputs are the six performance indices: energy consumption, kernel cracking, final moisture content, moisture removal rate, drying intensity and water mass removal rate. The inputs' four drying parameters -ice layer thickness (RLT), hot airflow rate (HAR), hot air temperature (THA) and drying time (TD)- are the inputs of the neural network. The optimal model is a four-layered backpropagation neural network, with 8 and 5 neurons in the first and the second hidden layers, respectively. The effectiveness of the proposed model is demonstrated using experimental data. The mean relative error varied from 2.0% to 8.3% for 6 predictions with average of 4.4%. Using a multiple-objective programming for optimization of the drying parameters, the optimal values are: RLT=66 cm, HAR=0.30 m/s, THA=93/spl deg/C and TD=23 min.