Hybrid Model Predictive Control of a Solar Air Conditioning Plant

This paper describes the development and experimental validation of a multi-layer hybrid controller for optimizing energy management in a solar air conditioning plant. The hybrid nature of the process is due to its multi-mode structure: the refrigeration circuit can be fed by flat solar collectors, a storage system, by an auxiliary gas heater, or by a combination of them. The selection of the operating mode is obtained by switching electrovalves, pumps, and three-way mixing valves. The proposed multi-layer hybrid controller consists of a high-level supervisor that decides on-line the optimal operating mode through a hybrid model predictive control strategy, a static lower-level controller defining proper set-points for the chosen mode, and existing standard low-level controllers that ensure robust tracking of such set-points. The overall controller was designed in Matlab/Simulink using the Hybrid Toolbox, and then tested experimentally a real process, showing the effectiveness of the approach.