Dynamic mode decomposition for modeling the UI dependence in fuel cells

Modeling and prediction of fuel cell dynamics is a challenging task and has recently gained importance. However, the underlying physical processes are often complex, and the dynamics is determined by a large number of parameters that are not completely known. We present results of a model based on dynamic mode decomposition (DMD) for proton exchange membrane fuel cells (PEMFC). Our simulation results show that the proposed prognostic strategy on the basis of DMD allows both confirmation of experimental observations and prediction of future behavior. In particular, the transition to a regime characterized by degradation can be monitored and recognized in advance.