A Bayesian Approach for Spherical Harmonic Expansion Identification: Application to Magnetostatic Field Created by a Power Circuitry

Olivier Pinaud (1) , Olivier Chadebec (1) , Laure-Line Rouve (1) , Jean-Louis Coulomb (1) , Jean-Michel Guichon (1) , A. Vassilev (2)

1 G2ELab - Laboratoire de Génie Electrique de Grenoble
2 CEA-LETI - Commissariat à l'énergie atomique et aux énergies alternatives - Laboratoire d'Electronique et de Technologie de l'Information

This paper deals with the use of the Bayesian approach to inverse an underdetermined magnetostatic problem based on spherical harmonic expansion. Identification of the spherical harmonic coefficients is helped thanks to some a priori information. This information comes from a numerical model statistically studied to define an average-state vector and a covariance matrix. The whole approach is applied for the study of the magnetostatic field inside an electric vehicle, created by its power circuitry. It demonstrates the strength of merging a priori information and measured information in order to obtain an efficient identification of magnetic sources created by a complex set of conductors.

Mots clés: Bayes theorem; inverse problem theory; magnetostatics; random variables propagation; spherical harmonics

Domaines: Energie électrique

Olivier Pinaud, Olivier Chadebec, Laure-Line Rouve, Jean-Louis Coulomb, Jean-Michel Guichon, et al.. A Bayesian Approach for Spherical Harmonic Expansion Identification: Application to Magnetostatic Field Created by a Power Circuitry. IEEE Transactions on Electromagnetic Compatibility, 2015, 57 (6), pp.1501-1509. ⟨10.1109/TEMC.2015.2458353⟩. ⟨hal-01261250⟩