Abstract
Hydrogen fuel cells are important devices for hydrogen energy utilization, and assembly clamping pressure can affect cell performance through the coupled effects of interfacial contact conditions, heat transfer, and mass transport. To address the high computational cost of conventional high-fidelity CFD-FEA coupled models and their limited applicability to rapid parameter scanning and optimization, this study constructs a dataset containing 81 operating cases based on previous clamping-pressure coupled simulations. Three artificial intelligence surrogate models, namely radial basis function neural network, support vector regression, and Gaussian process regression, are trained and further coupled with a genetic algorithm to enable rapid prediction and optimization of power density. The results indicate that Gaussian process regression exhibits better overall performance for the small-sample, low-dimensional dataset, with a training time of approximately 0.4 s. The optimization results suggest that the clamping pressure and operating voltage ranges corresponding to high power density are 0.8∼1.2 MPa and 0.61∼0.63 V, respectively. Within these ranges, the single-cell output power density can be maintained at no less than 0.78 W·cm−2. This study provides a reference for PEMFC assembly pressure setting and operating parameter optimization.
| Translated title of the contribution | Study on the Impact of Hydrogen Fuel Cell Clamping Pressure on Cell Performance Based on Artificial Intelligence Methods |
|---|---|
| Original language | Chinese (Traditional) |
| Pages (from-to) | 3574-3580 |
| Number of pages | 7 |
| Journal | Kung Cheng Je Wu Li Hsueh Pao/Journal of Engineering Thermophysics |
| Volume | 47 |
| Issue number | 10 |
| State | Published - Oct 2026 |
| Externally published | Yes |
Fingerprint
Dive into the research topics of 'Study on the Impact of Hydrogen Fuel Cell Clamping Pressure on Cell Performance Based on Artificial Intelligence Methods'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver