Dynamic optimization of water temperature for maximizing leaf water content of tomato in hydroponics using an intelligent control technique

In this study, the dynamic optimization of water temperature, which maximizes the leaf water content of tomatoes in hydroponics, was conducted using neural networks and genetic algorithms. The leaf water content was estimated from leaf thickness in a continuous and non-destructive manner using an eddy current-type displacement sensor. Identification of model mechanisms was achieved and a dynamic model was built using neural networks. A three-layered neural network allowed such a complex system to be successfully identified and generated the model. Next, the control process was divided into six steps and the optimal six-step set points for water temperature that maximizes the leaf water content were examined by simulating the identified neural-network model, using genetic algorithms. The length of the control process was 10 hours, and each step took 90 minutes. The optimal 6-step set points for the water temperature were 40→10→40→10→40→38°C, under the constraint of a fixed water temperature from 10 to 40°C. From simulation, it was confirmed that this operation is effective in maximizing the leaf water content of the tomato. Finally, this operation was applied to a real system. The resulting leaf water content in this operation was about 1.15 times larger than that in a conventional control. It is suggested that this control technique is useful for promoting water uptake of the root and associated higher leaf water content during a short-term period within a day.