Description
Date depot: 13 juin 2022
Titre: Curve characterization and intelligent querying of an electrical consumption database
Directeur de thèse:
Themis PALPANAS (LIPADE)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Intelligence artificielle
Resumé: We nowadays record increasingly large volumes of time-indexed measurement data (time series) such as electricity production and consumption measurements. These time series are stored in large databases with associated context data: electricity usage, customer characteristics, consumption locations, etc. These databases contain a lot of missing information such as certain customer characteristics (e.g. family composition). Moreover, some information on the curves is rarely filled in but can be detected by experts (activity rhythm, presence/absence, EV charging) by visualizing the curve. However, it is impossible for these experts to label a large number of curves. One of the objectives of the thesis is to be able to characterize the uses and the behavior of the customers (individuals or companies) from their curve by an AI which would compare the labeled curves or would seek patterns characterizing the appliances or the activity periods. Another purpose would be to enrich the possibility of querying electricity consumption data by returning an estimate and its associated confidence interval to take into account the contextual data that are not filled in.
Doctorant.e: Petralia Adrien