Exploring the Capabilities of Adaptive Model Predictive Control in Irrigation Systems

This work addresses the problem associated with the variability of the parameters involved in irrigation control systems for real crops. Factors such as soil compaction, climatic variability or phenological state of the crops, among others, significantly influence the dynamics of these systems, challenging the implementation of model-based controllers in real use cases. In this context, an Adaptive Model Predictive Control scheme is proposed, which makes it possible to update the model by employing a recursive system identification. A comparison with a conventional predictive controller employing a constant model is made. The study is based on models identified from data collected in a production farm in Seville, Spain. The validation of the proposed strategy and the comparison between the adaptive MPC and the conventional MPC are performed by means of simulations. The results demonstrate the potential applicability and effectiveness of Adaptive MPC in real farming conditions.