2D and 3D fault basis for fuel cell diagnosis by external magnetic field measurements

Abstract

An original approach used for the identification of faults in fuel cell stacks is presented. It is based on the 3D reconstruction of the current density from external magnetic field measurements which is an ill-posed magnetostatic linear inverse problem. A suitable and original current density and magnetic field basis are proposed in order to define both local and global faults on a fuel cell stack. The inverse problem is regularized by truncated singular value decomposition (SVD) to ensure the uniqueness of the solution.