A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Automatic generation of multi-zone RC models using smart thermostat data from homes
Abstract An automated methodology that generates sufficiently accurate building models is essential for both the adoption of advanced control strategies like Model Predictive Control, and the estimation of building energy flexibility required for building-grid interaction. This paper presents such a methodology for the generation of multi-zone resistance-capacitance (RC) thermal network models of residential buildings. The models are calibrated using smart thermostat data and predict the indoor air temperature up to 24 h ahead. The methodology starts with a very simple model and iteratively adds one parameter at a time; the parameter that increases the quality of the model the most, as measured with the Bayesian Information Criterion. When the quality of the model cannot be improved any further, the procedure is reversed to delete any redundant parameters. The algorithm concludes when neither adding nor removing a parameter increases the model quality. The algorithm is applied on measured data from an experimental facility, a bungalow house in Québec, Canada. The resulting model can accurately predict all 9 zone temperatures 24 h in advance with an RMSE of 0.44 °C. An examination of the estimated parameters shows that they reflect the layout of the house, previously unknown to the methodology.
Automatic generation of multi-zone RC models using smart thermostat data from homes
Abstract An automated methodology that generates sufficiently accurate building models is essential for both the adoption of advanced control strategies like Model Predictive Control, and the estimation of building energy flexibility required for building-grid interaction. This paper presents such a methodology for the generation of multi-zone resistance-capacitance (RC) thermal network models of residential buildings. The models are calibrated using smart thermostat data and predict the indoor air temperature up to 24 h ahead. The methodology starts with a very simple model and iteratively adds one parameter at a time; the parameter that increases the quality of the model the most, as measured with the Bayesian Information Criterion. When the quality of the model cannot be improved any further, the procedure is reversed to delete any redundant parameters. The algorithm concludes when neither adding nor removing a parameter increases the model quality. The algorithm is applied on measured data from an experimental facility, a bungalow house in Québec, Canada. The resulting model can accurately predict all 9 zone temperatures 24 h in advance with an RMSE of 0.44 °C. An examination of the estimated parameters shows that they reflect the layout of the house, previously unknown to the methodology.
Automatic generation of multi-zone RC models using smart thermostat data from homes
Vallianos, Charalampos (author) / Athienitis, Andreas (author) / Delcroix, Benoit (author)
Energy and Buildings ; 277
2022-10-10
Article (Journal)
Electronic Resource
English
| Taylor & Francis Verlag | 2022
Effect of thermostat and window opening occupant behavior models on energy use in homes
| Online Contents | 2014