A fast implementation of coalitional model predictive controllers based on machine learning: Application to solar power plants

This article proposes a real-time implementation of distributed model predictive controllers to maximize the thermal energy generated by parabolic trough collector fields. For this control strategy, we consider that each loop solar field is individually managed controller, which can form coalition with other attain its local goals while contributing overall objective. The formation coalitions based on market-based mechanism in heat transfer fluid traded. To relieve computational burden online, propose learning-based approach approximates optimization problems so controller be applied real time. Finally, simulations 100-loop are used assess coalitional strategy neural networks comparison control. results show provides reduction computing time up 99.74% and minimal performance compared as baseline.