A novel pre-diagnosis method for health status of proton exchange membrane fuel cell stack based on entropy algorithms

Effective and accurate cell health status diagnosis is key to ensuring the stable operation of the fuel cell stack. The reliability of the current voltage value-based method is challenging due to the solid time-varying nature of fuel cells. This paper utilizes modified Shannon entropy to propose a novel method for fuel cell health status evaluation and pre-diagnosis. It is revealed that fuel cell health status can be effectively characterized by quantifying the voltage fluctuation degree using modified Shannon entropy. Furthermore, its sensitivity, universality, and reliability are verified by different types of experimental data, including extreme operating conditions, membrane electrode assembly's severe inconsistent aging, and unreasonable structures. Then, an abnormal coefficient considering the stack inconsistency is proposed utilizing the entropy combined with the Z-score method and can diagnose in-stack abnormal cells in advance based only on timing voltage. Further, the fuel cell's abnormality level can be determined in real time according to the established three-level health status management strategy. Corresponding treatments are recommended. Finally, the method's application prospect in practical systems such as vehicles and big data platforms is explored due to the small computation and easy implementation, which builds a foundation for the future fuel cell health management system.