A Hybrid Method for Performance Degradation Probability Prediction of Proton Exchange Membrane Fuel Cell

Area of Science:

• Energy Systems
• Materials Science
• Artificial Intelligence

Background:

• Proton exchange membrane fuel cells (PEMFCs) offer significant potential as clean energy sources.
• However, their widespread adoption is hindered by issues related to short operational lifespan and high maintenance expenses.
• Effective performance degradation prediction is crucial for mitigating these challenges.

Purpose of the Study:

• To develop a novel hybrid method for accurately predicting the performance degradation of PEMFCs.
• To enhance the reliability and reduce the maintenance costs associated with PEMFC technology.
• To provide a robust framework for understanding and forecasting PEMFC aging patterns.

Main Methods:

• A Wiener process model was employed to characterize the inherent randomness in PEMFC degradation.
• The unscented Kalman filter algorithm was utilized for real-time estimation of the aging factor's degradation state from voltage monitoring.
• A transformer network, enhanced with Monte Carlo dropout, was implemented for predicting future degradation states and quantifying prediction uncertainty via confidence intervals.

Main Results:

• The proposed hybrid method demonstrated effectiveness in predicting PEMFC performance degradation.
• The integration of Wiener process, unscented Kalman filter, and transformer with Monte Carlo dropout provided accurate degradation state estimations.
• The method successfully captured data characteristics and fluctuations of the aging factor, offering quantified uncertainty in predictions.

Conclusions:

• The novel hybrid method offers a superior approach to PEMFC performance degradation prediction compared to existing techniques.
• This predictive capability is vital for extending PEMFC lifespan and reducing operational costs.
• The study validates the effectiveness and advantages of the proposed method on experimental datasets, paving the way for improved PEMFC reliability.