Application of intelligent control based on fuzzy logic to the optimal control for accelerating calcium ion uptake of tomato plants in hydroponics

在水培番茄栽培中，由于缺钙会引起番茄的许多生理障碍，如果实的花梢腐烂等。在番茄栽培中，为了获得更好的果实品质，必须对水培系统进行有效的控制。然而，控制系统的特点是复杂性和不确定性。本研究将模糊逻辑、神经网络及遗传算法等智能控制技术应用于番茄水培栽培的优化控制。(1)在番茄栽培中，在生殖生长过程中观察到显著的钙离子吸收。因此，番茄果实缺钙可能是由于钙离子向果实的运输量较低而不是钙离子吸收量较低所致。(2)将基于模糊逻辑的反馈控制与基于神经网络的前馈控制相结合，提出了一种新的自适应控制方法，有效地解决了水培营养液的时变控制问题。应用模糊逻辑增加了控制性能的灵活性和平滑性。此外，应用基于神经网络的逆系统模型，获得了更快速和自适应的控制性能。(3)将神经网络与遗传算法相结合的最优控制技术用于植物净光合速率的优化。该方法首先利用神经网络对间歇供液的净光合速率进行辨识，然后利用遗传算法对模型进行模拟，确定间歇供液的4步排液和供液时间的最佳值。神经网络的使用使得识别净光合速率等复杂系统成为可能。此外，遗传算法使我们能够快速搜索最优值。

There are many physiological disorders, e.g.blossom-end rot of fruits, caused by calcium deficiency in a hydroponic tomato cultivation. In order to obtain better quality of fruits in a tomato cultivation, effective control of the hydroponic system is neccessay. However, the control system is characterized by complexity and fuzziness. In this study, intelligent control techniques based on fuzzy logic, neural networks and genetic algorithm are developed and then applied to the optimal control of hydroponic tomato cultivation.(1) In the tomato cultivation, marked calcium ion uptake was observed during the reproductive growth. So, the calcium deficiency in tomato fruits seems to be caused by rather low calcium ion transportaion to the fruits than low calcium ion uptake.(2) A newly developed adaptive control technique combining with the feedback control based on fuzzy logic and the feedforward control based on neural networks was effective for the time-varying control problem of the hydroponic nutrient solution. Applying the fuzzy logic increased flexibility and smoothness of the control performance. Furthermore, applying the inverse system model based on neural networks, more speedy and adaptive control performances were obtained.(3) A newly developed optimal control technique combining with neural networks and genetic algorithms was useful for optimization of net photosynthetic rate of a plant. In the method, the net photosynthetic rate to intermittent solution supply is at first identified by using neural networks and then optimal values of 4-step drainage and supply times for intermittent solution supply are determined through the model simulation by using genetic algorithms. The use of neural networks made it possible to identify such complex systems as net photosynthetic rate. Furthermore, the genetic algorithms enable us to quickly search the optimal values.