Accurate modelling of the drying process of agricultural material is an essential step toward the development of advanced drying monitoring and control systems. In this study, Artificial Neural Networks (ANNs), Adaptive Neuro-Fuzzy Inference Systems (ANFIS) and mathematical modelling were employed to predict the moisture ratio of paddy during the drying process in a combined hot air IR-assisted dryer in fixed and vibratory bed modes. The grains were dried as thin layers at three infrared power levels of 50, 100 and 150 W and temperature levels of 40, 50 and 60 °C. The simulation results proved that The ANN model with 3-18-16-1 topology, LM training function, logarithm sigmoid transfer function in the first, and tangent sigmoid transfer function in the second hidden layer was the most accurate model. The R2, RMSE and MAE values for the ANN model in the fixed-bed mode were 99.92%, 0.0037 and 0.0026, respectively. These values for vibratory mode were 99.94%, 0.0030 and 0.0021, respectively. It was concluded that ANN and ANFIS models were more appropriate tools than mathematical models for predicting the hot air-IR drying kinetics of paddy.
artificial intelligence, drying, infrared, modeling, paddy
Rahmanian-Koushkaki, H., Bakhshipour, A., Nourmohammadi-Moghadami, A. (2023): Prediction of Paddy Drying Kinetics in an IR Assisted Hot-Air Dryer Using ANN and ANFIS. Scientia Agriculturae Bohemica, 54, 47-62. DOI: 10.7160/sab.2023.540206