Data-driven Iterative Learning for Model Predictive Control of Heating Systems*

Combining iterative learning with model predictive controllers is reasonable in applications where approximative models for the systems dynamics are available and relevant disturbances are repetitive, e.g. the outside temperature and the heat demand for heating systems. This paper shows how this combined control concept can be designed with a data-driven learning part, because for a rising number of application, signal histories are stored in databases. Simulation results for a heating system of a non-residential building are presented, which show the applicability of the approach.