Building Heterogeneous Models at Runtime to Detect Faults in Ambient-Intelligent Environments
Christophe Jacquet, Ahmed Mohamed, Frédéric Boulanger, Cécile Hardebolle, Yacine Bellik · 2013
Abstract. This paper introduces an approach for fault detection in ambient-intelligent environments. It proposes to compute predictions for sensor values, to be compared with actual values. As ambient environ-ments are highly dynamic, one cannot pre-determine a prediction method. Therefore, our approach relies on (a) the modeling of sensors, actuators and physical effects that link them, and (b) the automatic construction at run-time of a heterogeneous prediction model. The prediction model can then be executed on a heterogeneous modeling platform such as ModHel’X, which yields predicted sensor values.