|Budget Amount *help
¥2,500,000 (Direct Cost : ¥2,500,000)
Fiscal Year 1997 : ¥700,000 (Direct Cost : ¥700,000)
Fiscal Year 1996 : ¥1,800,000 (Direct Cost : ¥1,800,000)
In this study, new techniques for indentification and optimal controle of fruit-storage process using fuzzy logic, neural networks and genetic algorithms were developed, aiming at the qualitative improvement of fruit during storage.
1) Fruit responses (water loss and skin color), as affected by environmental factors (temperature and relative humidity), during stroage were investigated using sensors. They showed dynamic changes against the step inputs of the environmental changes. The relationship between the water loss and the environment is characterized by linear while the relationship between the skin color and the environment is give by strong nonlinearity. So, it is found that they should be treated as dynamic and nonlinear systems.
2) A three-layr neural network was effective for both indentification of the linear system such as a water loss and the nonlinear system such as a skin color (ripening). On the other hand, genetic algorithms were shown to be very effective to find the io
ta-steps of optimal setpoints on temperature and relative humidity through simulation of the identified neural-network model.
3) A new fuzzy control technique, which efficiently selects optimal membership functions and control rules by using neural networks and genetic algorithms, was developed and then applied to the control of relative humidity in a fruit-storage house. The control input is the on-off of ventilation. The control aim is to maintain the relative humidity in the storage house at the desired value through the on-off control of ventilation by the fuzzy control. Results show that this control technique allowed optimal membership functions and control rules to be successfully determined and its control performance was superior to the conventional control.
4) Finally, an optimal pattern of heat treatment for tomatoes during storage was investigated based on their surface color, using an intelligent control technique consisting of neural networks and genetic algorithms. Two types of optimal heat treatments were obtained. One was the single application of heat, which is similar to the conventional type, and the other was intermittent application, given periodically. Finally, the two optimal treatments were applied to an actual system. The result showed that they gave better results on ripening than continuous cooling. Less