Model Predictive Control for Building Energy Reduction and Temperature Regulation

Building climate control mechanisms account for more than 50% of the overall residential and commercial sector energy usage. Other than undertaking complementary green building design procedure to cut down the operational cost, optimal control of air-conditioning and mechanism ventilation (ACMV) systems in existing buildings is mutually important. While current building manage systems (BMS) usually operates with proportional-integral controllers to maintain constant component set points, there is no supervisory optimization for overall system operation under various conditions. In this paper, we propose a model predictive control (MPC)-based optimal temperature controller suitable for on-line optimization for smart buildings equipped with sensors. The proposed MPC controller integrates building thermodynamics, occupancy data, weather forecast data, as well as ACMV component models for minimizing energy consumption as well as stabilizing building temperature. To ensure feasibility during real-time operation, the above mentioned optimization is further decoupled into two sub-optimizations, dealing with system thermodynamics and component power consumption characteristics separately. In the simulation studies, the proposed MPC controller is able to achieve as much as 18.2% energy saving with different temperature regulation settings.