Fault Identification on a Fuel Cell by 3-D Current Density Reconstruction From External Magnetic Field Measurements

An original approach of fuel cell diagnosis is presented. It is based on the solving of an inverse linear problem linking the magnetic field signature outside of the fuel cell to the current density distribution inside. The searched solution is a linear combination of conservative current distribution obtained by a set of electrokinetic problems solved by a finite face element method. As the problem is ill-posed, the solution is stabilized using a truncated singular value decomposition. The approach is validated to reconstruct a 3-D current density distribution in a stack simulator and in a fuel cell stack operating in the laboratory conditions.