A selective hybrid stochastic strategy for fuel-cell multi-parameter identification

The in situ identification of fuel-cell material parameters is crucial both for guiding the research for advanced functionalized materials and for fitting multiphysics models, which can be used in fuel cell performance evaluation and optimization. However, this identification still remains challenging when dealing with direct measurements. This paper presents a method for achieving this aim by stochastic optimization. Such techniques have been applied to the analysis of fuel cells for ten years, but typically to specific problems and by means of semi-empirical models, with an increased number of articles published in the last years. We present an original formulation that makes use of an accurate zero-dimensional multi-physical model of a polymer electrolyte membrane fuel cell and of two cooperating stochastic algorithms, particle swarm optimization and differential evolution, to extract multiple material parameters (exchange current density, mass transfer coefficient, diffusivity, conductivity, activation barriers …) from the experimental data of polarization curves (i.e. in situ measurements) under some controlled temperature, gas back pressure and humidification. The method is suitable for application in other fields where fitting of multiphysics nonlinear models is involved.