Modèle polynomial par morceaux muni de transitions régulières - Application à la modélisation de signaux transitoires électriques

Matthieu Sanquer (1) , Florent Chatelain (1) , Mabrouka El-Guedri (2) , Nadine Martin (1)

1 GIPSA-SAIGA - GIPSA - Signal et Automatique pour la surveillance, le diagnostic et la biomécanique
2 Electicité de France R&D

A smooth transition model is introduced and studied. Such a model extend piecewise regression ones by introducing smooth transition functions achieving transition from a segment to another in the neighborhood of each rupture. The joint estimation of the parameters and of the number of segment is an ill-posed problem. A regularization is performed through a hierarchical Bayesian framework. As standard Bayesian estimates can not be computed analytically, a reversible-jump MCMC algorithm is derived to sample the parameters according to their posterior distribution. This method is applied to the modeling of real-world electrical transient.

Domaines: Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Matthieu Sanquer, Florent Chatelain, Mabrouka El-Guedri, Nadine Martin. Modèle polynomial par morceaux muni de transitions régulières - Application à la modélisation de signaux transitoires électriques. GRETSI 2011 - XXIIIème Colloque francophone de traitement du signal et des images, Sep 2011, Bordeaux, France. pp.s/n. ⟨hal-00609204⟩