Evaluation of Thin-Layer Drying Models and Artificial Neural Networks for Describing Drying Kinetics of Canola Seed in a Heat Pump Assisted Fluidized Bed Dryer

In this study, drying characteristics of canola seeds were determined using heated ambient air at 40, 50 and 60 degrees C, relative humidity of 20, 40 and 60% and constant velocity of 3 m/s. To select a suitable drying curve, six thin-layer drying models were fitted to experimental data. The models were compared according to three statistical parameters: R-2, reduced chi-square (chi(2)) and root mean square error. Using some experimental data, an Artificial neural network model, trained by Feed Forward Back-Propagation algorithm, was developed to predictmoisture ratio values based on the three input variables. Different activation functions and several ruleswereusedtoassess percentage errorbetweenthedesired andpredicted values. Accordingtotheresults, theapproximation of diffusion drying model had better agreement with the drying data. The artificial neural network model was able to predict the moisture ratio quite well with R-2 of 0.9994. The predictedmean square error was obtained as 0.00012575.