A hierarchical Model-based Predictive Control strategy for building heating systems

A hierarchical Model Predictive Control (MPC) strategy for optimally controlling the heating system of a building is developed to satisfy comfort constraints with reduced energy consumption. MPC based strategies can prevent unnecessary energy use by predicting the future heating requirement of the building and only supplying heat when necessary. In this paper, prediction error estimation methods are used to derive reduced order building models from historical data. A hierarchical structure is implemented, with interior-point algorithms employed to solve the resulting optimization problems. A simulation platform, designed to replicate the thermal dynamics and the heating system mechanisms of a real building is used to test the strategy. The MPC formulation is compared to the standard weather compensation scheme used in the building on which the simulation platform is based.