Performance of Long Short-Term Memory Networks for Modeling the Response of Plant Growth to Nutrient Solution Temperature in Hydroponic

Abstract

This study examines the development of an approach for modeling the response of plant growth to nutrient solution temperature in hydroponic cultivation in a dynamic system. Nutrient solution temperature is one of the essential manipulating factors for plant growth in hydroponic cultivation. Determining the optimal control strategy of nutrient solution temperature during cultivation could lead to maximize the growth of the plant. By identifying the process using a dynamic system, the optimal control strategy can be determined. However, developing a dynamic model of plant growth to nutrient solution temperature is not easy due to physiological behavior between them are quite complex and uncertain. We propose the long short-term memory (LSTM) networks to identify and develop a model of dynamic characteristics of plant growth as affected by the nutrient solution temperature. Chili pepper plants were used to obtain time-series data of plant growth, with five different types of dynamic nutrient solution temperature patterns for system identification. The results showed that the proposed LSTM model provides promising performance in predicting the response of plant growth to nutrient solution temperature in hydroponic cultivation.