Paper Conference

Proceedings of BSO Conference 2018: Fourth Conference of IBPSA-England


Learning About Error Terms In Energy Models By Bayesian Calibration

Kathrin Menberg, Yeonsook Heo, Ruchi Choudhary

Abstract: Recently there has been an increasing trend in the application of Bayesian methods for calibration of building energy models (BEM). While many of those studies take into account uncertainties from different error sources, the posterior hyper-parameter results have rarely been assessed in detail. In this study, we assess the robustness of the Kennedy & O’Hagan framework and the sensitivity of its results on input information. We uncover that a lack of inference for the model bias has a more pronounced effect on calibration results than for model parameters. The found dependency of the results on the calibration settings highlights the need for an appropriate assessment of the calibration results.
Pages: 543 - 550