Data-Driven Model Reduction of the Moving Boundary Heat Pump Dynamic Model

Heat pump systems have the potential to be used as controllable load to compensate for the uncertainties in modern power systems. The moving boundary model for heat pumps is known for its accuracy and acceptable computational burden. However, this model is sometimes impractical because it requires detailed information on the mechanical structure of the internal loops and the thermal state of the refrigerant. A data-driven method based on the cascaded wiener model is proposed in this paper to simplify the moving boundary model. The proposed model is developed from an observation that the dynamic behavior of the heat pump is relatively linear over a large range of operation states, while the static behavior is very nonlinear. Through comparison of simulation results, the proposed model has close accuracy to the moving boundary model and can be a viable alternative for control design purposes.