Model Identification of HVAC Systems for Office Buildings Considering Real Environment

Abstract

We propose two schemes for identifying the parameters of models for predicting temperatures in multiple zones of office buildings. Real environments are considered for model predictive control (MPC) used to energy-efficiently operate heating, ventilation, and air conditioning (HVAC) systems. In the first part of this paper, we describe a system model of an HVAC system and the conditions of real office buildings in terms of how to estimate internal heat gain (IHG) and reduce the negative effects of uncertainties, including measurement bias. The following part shows the two schemes for solving these challenges with the key idea that typical patterns of IHG and the degree of influence of uncertainties are known when focusing on office buildings. Furthermore, we evaluated the error in predicting temperature with data measured from a real office building. We achieved a mean absolute error (MAE) of 0.36-0.37°C for 1-day ahead prediction and improved the standard deviation of MAE (27.0%), which was used as robustness measure.