A Smart and Predictive Heating System Using Data Fusion Based on the Belief Theory
Résumé
This paper investigates the way to model and handle the contextual data uncertainties in order to design a smart heating system that reduces energy consumption. To achieve this, we propose in this paper a data fusion system which provides a trend hypothesis (contextual and near future prediction) with its associated belief. The data to be fused are essentially the occupant's habits, the weather forecast as well as the notion of thermal comfort associated to the occupant's activities. Since the data are uncertain, erroneous and heterogeneous, we propose a multilevel data fusion system based on the belief theory of Dempster-Shafer for data combination and the Transferable Belief Model (TBM) for making the decision which will be a challenge. Despite the data complexity, the simulations are very satisfactory in terms of reducing energy consumption.