Data-Driven Multi-fault Diagnosis for H2/O2 and H2/Air PEMFCs

Enhancing the Proton Exchange Membrane Fuel Cells’ systems (PEMFCs) life span and reliability is a sensible point for the FCs’ manufacturers. In this framework, the Health-Code project works contribute to improve the onboard monitoring and diagnostic tool for onboard state-of-health assessment to prevent improper operating conditions that can severely affect the stack performance. The diagnosed faulty conditions are the improper water management (drying and flooding), the reactants’ starvations (fuel and oxidant), and the fuel quality contaminations (poisoning). The developed methodologies are mainly based on the use of the Electrochemical Impedance Spectroscopy (EIS) measurements oriented to multi-fault detection purposes. Experimental activity was performed both on H2/O2 PEMFC and H2/Air PEMFC technologies. Experimental data are used for methods learning and validation, while the final tool is validated on board, directly on real systems. The developed data-driven approach is presented in this paper. Particularly, the procedure development related to the relevant features’ extraction and the algorithm learning are reported. Finally, the algorithm off-line validation results are presented.