Parameter estimation of unknown properties using transfer learning from virtual to existing buildings

This study proposes a transfer learning (TL)-based inverse modelling to identify unknown building properties. examines the from virtual buildings existing buildings, especially for identifying wall U-value, HVAC efficiency and lighting power density (LPD). For this purpose, synthetic data were generated simulation results of sampled EnergyPlus models, then we developed artificial neural network (ANN) models using data. By adopting TL, ANN transferred domain evaluated on 61 buildings. As result, relative improvements in CVRMSE achieved by against trained only with buildings’ 8.85%, 10.34% 15.73% nominal cooling COP, U-value LPD, respectively. Moreover, it is expected that use TL enables model be reusable another group improved performance reduced training time.