Recoverable fault behavior oriented diagnostic-prognostic hybrid framework for proton exchange membrane fuel cells

Area of Science:

• * Energy systems and sustainable technology.
• * Advanced materials and electrochemical devices.

Background:

• * Proton exchange membrane fuel cells (PEMFCs) offer high efficiency and environmental benefits but face durability challenges hindering commercialization.
• * Accurate prognostic techniques are vital for optimizing PEMFC operation and lifespan, yet recoverable faults often compromise their reliability.
• * Addressing interference from recoverable faults is critical for dependable PEMFC prognostics.

Purpose of the Study:

• * To develop a robust diagnostic-prognostic hybrid framework for PEMFCs.
• * To enhance the accuracy of remaining useful life (RUL) prediction by mitigating the impact of recoverable faults.
• * To improve the overall durability and operational stability of PEMFC systems.

Main Methods:

• * Implementation of a data-driven recoverable fault recognition model using data augmentation to handle data imbalance.
• * Development of a prognostic strategy-guided particle filter (PSG-PF) for degradation trend prediction and RUL estimation.
• * Validation using dynamic load aging experimental data from PEMFCs.

Main Results:

• * The recoverable fault recognition model achieved over 98% accuracy.
• * The hybrid framework demonstrated an average relative accuracy of 86% in RUL prediction after effectively mitigating recoverable fault impacts.
• * The system proved effective in complex operating conditions.

Conclusions:

• * The proposed hybrid framework reliably recognizes recoverable faults and provides credible RUL predictions for PEMFCs.
• * This approach significantly enhances the robustness of PEMFC prognostics.
• * The findings support longer operational lifetimes and more stable performance in fuel cell applications.