Identifying informative energy data in Bayesian calibration of building energy models

Abstract

Bayesian computation has received increasing attention in calibrating building energy models due to its flexibility and accuracy. However, there has been little research on how to determine informative energy data in Bayesian calibration in building energy models. Therefore, this study aims to determine and choose informative energy data using correlation analysis and hierarchical clustering method. A case study of retail building is used to demonstrate the proposed methods to infer four unknown input parameters using EnergyPlus program. The results indicate that the different combinations of energy data can provide various levels of accuracy in estimating unknown input variables in model calibration. This suggests that Bayesian computation is suitable for inferring the parameters when there are missing energy data that can be treated as uninformative output data. The proposed method can be also used to find the redundant information on energy data in order to improve computational efficiency in Bayesian calibration.