Abstract In this study, effect of osmotic dehydration and hot air-drying conditions on shrinkage, rehydration capacity, and moisture content of dried kiwifruit was investigated and artificial neural network and genetic algorithm were applied as an intelligent modeling system to predict these physicochemical properties. Kiwifruit slices were immersed in osmotic solutions (30%, 40%, 50%, and 60%) at different temperatures (20, 40, and 60 °C) and were dried at 60, 70, and 80 °C for 5, 6, and 7 h. The results showed that increasing drying time and temperature caused an increase in shrinkage, while rehydration capacity and moisture content were decreased. Neural network model with one hidden layer, four inputs (operating conditions) was developed to predict three outputs (shrinkage, rehydration capacity as a well as moisture content) and genetic algorithm was used to optimize network structure and learning parameters. Artificial neural network models were then tested against an independent dataset and results showed the ability of optimized intelligent model to estimate shrinkage, rehydration capacity, and moisture content with high correlation coefficients (0.94, 0.93, and 0.96, respectively). Moreover, sensitivity analysis indicated that the most sensitive input variable toward such predictions was drying time.

Keywords