Paper Conference

Proceedings of Building Simulation 2021: 17th Conference of IBPSA

     

Transfer learning based inverse modeling to identify unknown building properties

Yun-Dam Ko, Cheol-Soo Park
Seoul National University, Korea, Republic of (South Korea)

DOI: https://doi.org/10.26868/25222708.2021.30686
Abstract: This study proposes a transfer learning (TL)-based inverse approach to identify unknown building properties from monthly energy data. For this purpose, the artificial neural network models were developed from simulation results of sampled EnergyPlus models and then transferred to the domain of existing buildings. It is found that the TL models can be used to identify unknown factors in existing buildings. In addition, data collected from EnergyPlus simulation runs can improve the performance of a data-driven model when TL is adopted. It is expected that the use of TL enables the developed model to be reusable for another group of buildings with improved performance and reduced training time.
Keywords: Transfer learning, Building energy benchmarking, Clustering, Machine learning
Pages: 1896 - 1903
Paper:
bs2021_30686