Model predictive control of chilled water temperature for centrialized HVAC systems

This paper proposes a model predictive controller (MPC) to regulate the chilled water temperature of centralized Heating, Ventilation and Air Conditioning (HVAC) systems. Firstly, a linear parameter varying (LPV) model is built to represent the dynamics of the heat transfer between the chilled water and the refrigerant under different working conditions. Then, local linear models are identified at three chosen representative working conditions via local tests. A global LPV model is thereafter obtained by interpolating all local models along with transition process data. With the aid of the model, a MPC mechanism is implemented to adjust the compressor frequency in order to maintain the outlet chilled water temperature to a prescribed constant even in the presence of varying cooling load and external disturbance. Experimental tests on an actual HV AC system demonstrate the effectiveness and efficiency of the proposed scheme.