In-situ optimal control of nutrient solution for soilless cultivation

Soilless cultivation is one of main production modes in precision agriculture. Nutrient solution for soilless cultivation primarily consists of macroelements such as NO₃^-, H₂PO₄^-, K⁺, Ca²⁺, Mg²⁺, and SO₄^2-. In vegetable-nutrient solution system, there are complicated correlations between greenhouse vegetable growth and nutrient solution, thus it is a challenge to achieve optimal control of nutrient solution. In this work, based on Q-learning, we first propose an in-situ optimal control method of nutrient solution compositions for greenhouse vegetable. Instead of modeling the correlations between greenhouse vegetable growth and nutrient solution, this method searches for optimal control policy through systematic interaction with the environment. The effect of nutrient solution compositions on photosynthetic rate of greenhouse vegetable is experimentally investigated, and on this basis reward function is designed. The experimental results show our method is effective and practical.