Artificial-neural-network-based model predictive control to exploit energy flexibility in multi-energy systems comprising district cooling

District cooling systems (DCSs) belonging to multi-energy can be managed by model predictive controls (MPCs) designed reduce the amount of electrical energy collected from grid for backup when there is a temporal mismatch between demand and availability. In this paper, DCS recovering cold thermal liquid-to-compressed natural gas fuel station used in an 8-user residential neighborhood provide space summertime. neighborhood, system, including DCS, photovoltaic panels, based on variable-load air-to-water heat pumps. One user district was allowed manage its with MPC artificial neural network (ANN). By integrating ANN-based routine building simulation environment unlocking flexibility thermostatically controlled loads (TCLs) using variable setpoints, it possible consumption up −71% respect reference case rule-based control. This work highlights also importance ANN training process proper representation TCL model, which not trivial aspect taken into account data driven models.