Identifying whole-building heat loss coefficient from heterogeneous sensor data: An empirical survey of gray and black box approaches

Design of a short perturbation method for on-site estimation of a building envelope thermal performance

“…However, long equipment downtime may be problematic for practitioners and occupants. In addition, Baasch et al [17] found in a numerically based study that occupancy contributed greatly to the uncertainty of HTC estimations inferred from both grey-and black-box models.…”

“…Calculation R eq precisely the problem encountered by Baasch et al [17]. To alleviate this issue, Baasch et al chose to narrow their study down to only one first-and one secondorder model, without any model selection process, which makes interpretation of the results all the more difficult.…”

A large scale artificial dataset to evaluate the impact of on-board monitoring set-ups on the Heat Transfer Coefficient assessment

“…A step towards the thermal performance assessment on large scale was researched in the work of Baasch et al [23] who applied black and grey box modelling techniques on a large synthetical dataset. In the work, even though a simplified building stock was considered, the methods showed drawbacks in robustness and validation and they were not ready for application in a practical context.…”

Towards automated stochastic Grey-Box Model Calibration for Heat Transfer Coefficient Inference

“…A major constraint of RC model calibration is that they need fine tuning and expert knowledge [3,6]. In particular, for the RC model selection procedure to succeed at fitting a given dataset, the models in the set need to be robust and the number of models need to be sufficient to cover a wide variety of envelope dynamics and insulation levels [6]. However, the larger the model set, the more tedious it becomes to calibrate all models and then proceed to model selection.…”