Optimum strategies for the control of root-zone temperature to promote early-stage growth of chili pepper in soilless culture using an intelligent approach

In controlled environment agriculture it is crucial to control environmental factors optimally during cultivation in order to promote plant growth.
One of the essential manipulating factors for promoting plant growth is root-zone temperature.
In this study, the optimal strategies for root-zone temperature control in hydroponic cultivation were investigated for promoting the early-stage growth of chili pepper plants (Capsicum annuum L.). An optimal control strategy was examined using an intelligent control technique based on the plant speaking approach concept.
This technique consists of a neural network model to identify the dynamic responses of plant growth as affected by root-zone temperature and a genetic algorithm to search for the optimal set-point of root-zone temperature.
The experiment was conducted inside a controlled environment growth chamber during 60 days of observation.
A non-destructive and continuous plant weight measurement system based on a load cell was developed for measuring the dynamic response of plant growth as affected by the change of root-zone temperature.
Five data sets, which consisted of five dynamic responses of plant growth as affected by the dynamic root-zone temperature regimes within the range of 15-37°C, were used for system identification.
Through simulation, using an identified neural network model, the optimal control strategies of root-zone temperature that maximized plant growth were then estimated using a genetic algorithm.
The result showed that neural networks are useful in identifying the dynamic responses of plant growth; then, optimization using a genetic algorithm indicated that the dynamic control method with a daily control interval could offer the most effective way to promote the early-stage of plant growth of chili pepper.

III Asian Horticultural Congress – AHC2020

G.K. Aji, K. Hatou, T. Morimoto

hydroponic, dynamic modeling, system identification, dynamic optimization, neural networks, NARX, LSTM, genetic algorithm

https://doi.org/10.17660/ActaHortic.2021.1312.26