Diagnosis of a fuel cell stack using electrochemical impedance spectroscopy and Bayesian Networks

In the first part of the paper, a novel architecture of impedance spectrometer is presented. The instrument is dedicated to the characterization and diagnostic of large fuel cell stacks operated in galvanostatic mode. It allows impedance measurements on cells located in the middle of the stack, where common mode potentials are usually too high for commercial devices. In the second part of the article, probabilistic methods (Bayesian Networks) are used to provide efficient diagnostic of a proton exchange membrane fuel cell stack. Experiments are performed on a twenty cell assembly using electrochemical impedance spectroscopy and from the test results, a Bayesian Network is proposed to ensure the diagnosis of the investigated fuel cell (study of drying and flooding phenomena). A maximum rate of good classification equal to 91% has been reached.