PREDICTION OF QUALITY INDICES DURING DRYING OF APPLES USING ARTIFICIAL NEURAL NETWORKS MODELS FOR PROCESS OPTIMIZATION

KARINA DI SCALA; GUSTAVO MESCHINO; ANTONIO VEGA-GÁLVEZ; JUDITH VERGARA; SARA ROURA; RODOLFO MASCHERONI

Valencia

Conferencia; International Conference on Food Innovation. FoodInnova, 2010.; 2010

Research Institute of Food Engineering for Development

Drying conditions generally affect several physico-chemical, nutritional and organoleptical food quality indices. The aim of this work is the application of Artificial Neural Network (ANN) models for predicting quality indices during convective drying of apples for process optimization. Multi-layer neural network models with two inputs: air temperature (40-80ºC) and air flow rate (0.5-1.5 m/s) were developed to estimate three outputs: colour (E), polyphenols (P) and water holding capacity (WHC). The Leave-one-out cross validation error was applied. A multiobjective Genetic algorithm (GA) was coupled with the configuration of neural network to find optimal drying conditions. Optimal values of 26.37 of E, 42.05 [mg GAE/100 g d. m.] of P and 51.61 [g retained water/100 g water] of WHC were found at 62.9 ºC and 1.0 m/s. Results indicated that the hybrid ANN?GA model could be effectively used not only to describe the quality indices of dehydrated apples but also to identify optimal drying conditions.