Artificial intelligence-assisted optimization and multiphase analysis of polygon PEM fuel cells

ABSTRACT

This study introduces innovative, optimized hexagonal and pentagonal PEM fuel cell models. The inlet pressure and temperature serve as input parameters, while power consumption and output power are objective parameters. The results of Computational Fluid Dynamics (CFD) analysis are then trained with deep neural networks and modeled using polynomial regression. Target functions are derived using the Response Surface Method (RSM) and optimized with the NSGA-II genetic algorithm. Compared to the base model, our optimized pentagonal and hexagonal PEM fuel cells significantly boost the output current density by 21.8% and 39.9%, respectively. Additionally, power consumption is lower: the pentagonal model uses 0.198%, and the hexagonal model uses 6.21% of the production power on average. Our proposed designs enhance PEM fuel cell performance by significantly boosting power production while minimizing power consumption.

KEYWORDS:

• PEM fuel cells
• artificial intelligence
• optimization techniques
• CFD analysis
• neural networks
• polygon
• NSGA-II

Disclosure statement

No potential conflict of interest was reported by the author(s).