Improving the prediction accuracy of building energy consumption using location of occupant - A case study

On the one hand, energy consumption forecasting in buildings is of great practical interest due to the large amount of energy that is consumed in buildings and therefore the big energy saving potential. Improving the prediction accuracy has attracted more and more attentions in recent years but still remains an open question. On the other hand, recent advances in technology has provided various economically affordable ways to obtain the location of the occupant. In this work, we focus on how to improve the prediction accuracy of building energy consumption using location of occupant. Three major contributions have been made. First, we formulate the energy consumption prediction problems as Markov decision processes. Second, we develop a platform including a lab, an apartment, and one occupant. The location of the occupant as well as the energy consumption in the lab and the apartment are monitored in the platform. Third, we show that the prediction accuracies of the energy consumption of both the buildings and the occupant can be improved using the location of the occupant. We hope that this work sheds some light on improving the energy efficiency of buildings in the near future.