Time-dependent Outlier Detection by means of h-mode depth and dynamic time warping
Résumé
This paper deals with the problem of finding outliers, i.e. data that differ distinctly from other elements of the considered dataset, when they belong to functional infinite-dimensional vector spaces. Functional data are widely present in the industry and may originate from physical measurements or numerical simulations. The automatic identification of outliers in this case can help to ensure the quality of a dataset (trimming), to validate the results of industrial simulation codes, or to detect specific phenomena or anomalies. In this paper, we focus on data originated from expensive simulation codes, such as nuclear thermal-hydraulic simulators, in order to take into account the realistic case where only a limited quantity of information about the studied process is available. A detection methodology based on the use of features like the h-mode depth and the Dynamic Time Warping in order to evaluate the outlyingness both in the magnitude and shape sense is proposed. Finally, its results are evaluated on the basis of some theoretical examples and a thermal-hydraulic application case.
Fichier principal
SpIssue_Journal_Quality_Technology_DoubleSpaced.pdf (758.07 Ko)
Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...