In this research, green bell pepper was dried in a pilot plant fluidized bed dryer equipped with a heat pump humidifier using three temperatures of 40, 50 and 60C and two airflow velocities of 2 and 3 m/s in constant air moisture. Three modeling methods including nonlinear regression technique, Fuzzy Logic and Artificial Neural Networks were applied to investigate drying kinetics for the sample. Among the mathematical models, Midilli model with R = 0.9998 and root mean square error (RMSE) = 0.00451 showed the best fit with experimental data. Feed‐Forward‐Back‐Propagation network with Levenberg–Marquardt training algorithm, hyperbolic tangent sigmoid transfer function, training cycle of 1,000 epoch and 2‐5‐1 topology, deserving R = 0.99828 and mean square error (MSE) = 5.5E‐05, was determined as the best neural model. Overall, Neural Networks method was much more precise than two other methods in prediction of drying kinetics and control of drying parameters for green bell pepper.
Practical Applications
This article deals with different modeling approaches and their effectiveness and accuracy for predicting changes in the moisture ratio of green bell pepper enduring fluidized bed drying, which is one of the most concerning issues in food factories involved in drying fruits and vegetables. This research indicates that although efficiency of mathematical modeling, Fuzzy Logic controls and Artificial Neural Networks (ANNs) were all acceptable, the modern prediction methods of Fuzzy Logic and especially ANNs were more productive and precise. Besides, this report compares our findings with previous ones carried out with the view of predicting moisture quotients of other food crops during miscellaneous drying procedures.