Nonlinear model predictive control for a heating and cooling system of a low-energy office building

Abstract

Model predictive control (MPC) is highly suitable for building heating and cooling control because it exploits disturbance predictions, obeys constraints, and enables optimal building operation in terms of user comfort and energy efficiency. This work presents a highly efficient nonlinear modular MPC concept. It optimally controls both, heating and cooling activities in a low-energy office building simultaneously. Relevant system nonlinearities are considered through a nonlinear prediction model, an LTI MPC optimization step, and an efficient mixed-integer mapping to setpoint temperatures in the building. The involved optimization problems are efficiently solvable and enable realtime control, and the controller structure allows for retrofitting and can directly be incorporated into the existing building control infrastructure. A clear formulation of thermal user comfort and energy efficiency allows straightforward tuning. Excellent control performance and robustness are observed in detailed co-simulation studies, significantly outperforming a classical rule-based reference control law. Possible approaches to analyze robust stability of the controlled system are discussed and related to results of robust TS-fuzzy system analysis.