DYNAMIC OPTIMIZATION USING NEURAL NETWORKS AND GENETIC ALGORITHMS FOR TOMATO COOL STORAGE TO MINIMIZE WATER LOSS

Abstract

In this study, dynamic optimization of heat treatment for reducing the water loss in fruit during storage wasinvestigated using intelligent approaches. Over a temperature range from 15.C to 40.C, the control process was divided intol steps (l = 6). The dynamic change in the rate of water loss as affected by temperature was first identified using neuralnetworks, and then the optimal combination of the lstep setpoints for temperature that minimized the rate of water loss wassearched for through simulation of the identified model using genetic algorithms. Two types of optimal values, a singleapplication of heat stress and a double application of heat stress, were obtained under the range of 15.C < T < 40.C. Thelength of each step was 24 h. The former treatment is useful for shortterm storage, and the latter is useful for comparativelylongterm storage. With the single heat treatment, the temperature first rises to the highest level (40.C), which is maintainedover a period of 24 h, and then suddenly drops to the lowest level (15.C). In particular, the sudden drop in temperature fromthe highest level to the lowest level provided lower values of the rate of water loss than maintaining the temperature constantlyat the lowest level throughout the control process. These results suggest that application of heat stress to fruit is effective inmaintaining freshness of fruit during storage.