Prediction of Moisture Content for Papaya Fruit During The Drying Process Using Artificial Neural Networks Technique

Abstract

Non-destructive method for characterization of various crops during the harvesting process, significant progress had recently. The purpose
of this study was using artificial neural networks to predict the moisture content of papaya in the laboratory dryer with a drying of the Cabinet.
Temperature effect in three levels (40,50 and 60 degrees Celsius) and the thickness of the cut pieces (3,5 and 7 mm) on the moisture content
changes were studied. MLP with three layers back propagation neural network to model the drying process was designed with different learning
algorithms. Different topologies with different threshold functions were used to predict product moisture content. Results showed that the MLP
network with 1-9-3 structure and learning algorithm Levenberg-Marquardt with logarithmic sigmoid threshold function with different topologies
and learning algorithms provide better results. Coefficients R2 and RMSE for these networks were optimized, respectively, 0.9994 and 0.00708
Indicating the ability of artificial neural networks to model the drying of the product.