A Non-linear Model of Sensitivity Matrix for Electrical Capacitance Tomography

Gustavo Villares (1) , Laura Begon-Lours (1) , Cédric Margo (1) , Yacine Oussar (1) , Jérôme Lucas (1) , Stéphane Holé (1)

1 LPEM - Laboratoire de Physique et d'Etude des Matériaux (UMR 8213)

The estimation of the permittivity distribution within a given volume by Electrical Capacitive Tomography relies on a relationship between the measured signal and the permittivity distribution, known as the sensitivity matrix. We show that relative permittivity variations must be lower than 25% to obtain errors smaller than 10% with the model whereas the permittivity can be scaled by up to 3.6 for the same error range with the proposed non- model. The errors remain below 15% over the whole range of permittivity.

Domaines: Apprentissage [cs.LG]; Machine Learning [stat.ML]; Instrumentations et Détecteurs [physics.ins-det]

Gustavo Villares, Laura Begon-Lours, Cédric Margo, Yacine Oussar, Jérôme Lucas, et al.. A Non-linear Model of Sensitivity Matrix for Electrical Capacitance Tomography. 2012 Annual Meeting of the Electrostatics Society of America, Jun 2012, Canada. pp.00 - 00. ⟨hal-00927194⟩