Optimization for the Model Predictive Control of Building HVAC System and Experimental Verification

Abstract

This article presents an optimized prediction model of building dynamic HVAC system load, which simplifies the input parameters of the model while meeting the accuracy requirements of the prediction results. The model was established using the open-source Modelica-based building library, and the linear aggregation method was used to establish the model. A reduced-order model was developed, and the accuracy of the simplified and reduced-order models was verified. A control strategy was constructed using the indoor mean radiant temperature (MRT) aggregated from a simplified prediction model of HVAC system load as the target feedback parameter, and its feasibility was verified experimentally. It was found that the MRT adopted by the new control strategy can reflect the changes in outdoor air temperature and load in a timely manner; moreover, using this as a control parameter can significantly reduce the influence of load changes to maintain a stable indoor temperature. The control system is further simplified by the predictive model, which improves the engineering practicability by maintaining the control accuracy.