MODELING OF DRYING KINETIC OF PUMPKIN: PART II. ARTIFICIAL NEURAL APPROACH

In this research, thin-layer drying of pumpkin slices was simulated via a laboratory scale hot air dryer. The drying process was carried out at four different temperatures (65℃, 75℃, 85℃ and 95℃). Multilayer perceptron neural network (MLP) and radial basis function network (RBF) were implemented to forecast the moisture ratio and drying rate of samples during drying. Optimized artificial neural networks (ANNs) models were developed for MLP based on one hidden layers with topology 2-15-2 and 2-3-2 for moisture ratio and drying rate, respectively. In addition, RBF revealed the superlative results accompany with 30 nodes per first layer for both dying properties drying rate and moisture ratio. Thus, it can be concluded that MLP models gave better results than RBF models for monitoring the moisture ratio.

Keywords:

Pumpkin slices , Radial basis function , Multilayer perceptron