Post-prognostics decision making for a two-stacks fuel cell system based on a load-dependent deterioration model

Jian Zuo (1) , Catherine Cadet (1) , Z Li (2) , Christophe Bérenguer (1) , R Outbib (2)

1 GIPSA-SAFE - GIPSA - Safe, Controlled and Monitored Systems
2 LIS - Laboratoire d'Informatique et des Systèmes (LIS) (Marseille, Toulon)

Multi-stacks proton exchange membrane fuel cell (PEMFC) system has been applied to combined heat and power system (CHP), and serves as an alternative energy device due to its high efficiency and zero-emission. Owing to the limited durability and larger power supply demand, the management of multi-stacks PEMFC system to obtain a longer service time has received recently growing attention. From the prognostics and health management (PHM) point of view, a post-prognostics decision making for multi-stacks PEMFC system is addressed in this work. Firstly, a load-dependent stochastic deterioration model is proposed for PEMFC. The overall ohmic resistance is chosen as the health indicator of PEMFC. Then the resistance is modeled using a Gamma process whose shape parameter is taken as a function of the current load applied to the stack. Finally, for the post-prognostics decision-making phase, a decision probability-based load repartition criterion is built to identify the optimal load split between the two stacks. The decision probability is calculated based on the system lifetime results (EoL) in each decision step. The EoL results of the decision phase are further compared with the system EoL that calculated without the decision making strategy. The comparison result shows that extended service time can be achieved using the proposed post-prognostics decision-making method.

Mots clés: Fuel Cell; Prognostics; decision-making; load-dependent model

Domaines: Automatique / Robotique

Jian Zuo, Catherine Cadet, Z Li, Christophe Bérenguer, R Outbib. Post-prognostics decision making for a two-stacks fuel cell system based on a load-dependent deterioration model. PHME 2020 - European Conference of the Prognostics and Health management (PHM) Society, Jul 2020, Turin, Italy. pp.9. ⟨hal-02946950⟩