Abstract
A preliminary study was carried out to estimate the changes in inner quality indexes, such as total soluble solids (TSS), titratable acidity (TA), and aroma of apple in modified atmosphere packaging using a 3 layered backpropagation neural network model.
Fuji and Ohm apples were packed into polyethylene bags and stored at 3°C, 10°C and ordinary room temperature for 4-22 weeks. Eleven characteristic values of apple stored under the condition of modified atmosphere packaging, such as TSS, TA, skin color, stiffness, mass volume, CO2, O2 and aroma concentrations in the bag, etc. were measured by destructive and non-destructive methods at 14 day intervals during storage. The concentrations of CO2 and O2 in the bag, hunter lightness as well as hue angle values of external skin color and weight loss were measured by non-destructive methods, and these values were selected as input data of neural network for inner quality estimation. On the other hand, the measured values of TSS, TA and aroma concentrations were fed into the model as a target input.
As a result of simulations for TSS, TA and aroma estimations of stored apples, the averaged values of TSS, TA and aroma concentrations which were outputs of the model were nearly equal to those of averaged measured values by the destructive methods. Therefore, it was recognized that the neural network model was one of the effective supporting technologies for non-destructive estimation of inner quality of stored apples.
