Screening of the alterations in qualitative characteristics of grape under the impacts of storage and harvest times using artificial neural network

The tested model showed that high reliability based on the obtained inputs for achieved data for RMSE and correlation coefficient between the obtained experimental and predicted values. An enhancement in the storage time, reduced the pH value and flavor index (fruit maturation), but boosted the acidity value of the fruits. On the other hand retardation in the harvest time led to an increase in pH value, total soluble solids and dextrose contents as well as flavor index of the samples. Artificial neural network design has been applied to predict the process of alterations during storage time. Back propagation feed forward neural network with the arrangement of 5:8:2 with a high correlation coefficient value ((>)0.989) and low root mean square error value (< 0.0019) as well as sigmoid hyperbolic tangent activation function with Levenberg-Marquardt learning and learning cycle of 1000 were detected as the most reliable and appropriate neural network model in storage process.