Kinetic modeling of mass transfer during Kiwifruit osmotic dehydration operation by Artificial Neural Network

In the current research the artificial neural network models were used for predicting mass transfer kinetics of osmotically dehydrated kiwifruit. Osmotic dehydration operations were performed in sucrose solution with concentrations of 30, 40, 50 and 60% at temperatures of 20, 40 and 60°C for 30, 60, 90 and 120 minutes. Multi-layer neural network with 3 inputs (operating conditions) was developed to predict solid gain and water loss of osmotically dehydrated kiwifruits. It was found that artificial neural network model with 9 neurons in hidden layer gives the best fitting with the experimental data, which made it possible to predict solid gain and water loss with correlation coefficient of 0.93 and 0.99, respectively.