A variational guided metaheuristic framework based on AOA for parameter estimation in PEM fuel cells
Results in Engineering, cilt.30, 2026 (ESCI, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 30
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.rineng.2026.110878
- Dergi Adı: Results in Engineering
- Derginin Tarandığı İndeksler: Emerging Sources Citation Index (ESCI), Scopus
- Anahtar Kelimeler: Arithmetic optimization algorithm (AOA), Exploration and exploitation, Metaheuristic search algorithm (MSA), Parameter estimation, Proton exchange membrane fuel cell (PEMFC)
- Karadeniz Teknik Üniversitesi Adresli: Evet
Özet
Accurate parameter estimation of proton exchange membrane fuel cell (PEMFC) models is essential for optimizing fuel cell performance, supporting energy efficiency analysis through accurate electrochemical modeling, and enabling informed performance evaluation in low-carbon energy technologies. However, the underlying optimization problem is nonlinear, multimodal, and highly sensitive to initial conditions, posing challenges to traditional algorithms. This study introduces a novel metaheuristic framework, Variational Guided Arithmetic Optimization Algorithm (VGAOA), with three synergistic modifications to enhance the original Arithmetic Optimization Algorithm (AOA). First, the Diversity-Driven Population Seeding (DDPS) mechanism leverages chaotic mapping and oppositional learning to initialize a diverse population. Second, Tailored Influence Guidance (TIG) adaptively assigns individualized elite exemplars to solutions, ensuring decentralized learning and preventing premature convergence. Third, Dimension-Selective Peer Interaction (DSPI) enables adaptive, dimension-wise knowledge exchange for improved local exploitation. These innovations constitute a process innovation at the optimization search level that significantly strengthens the algorithm’s search dynamics. Extensive experiments on four real-world PEMFC datasets (Temasek, NedStack PS6, Horizon H-12 and single cell PEMFC) demonstrate that VGAOA outperforms state-of-the-art optimizers in convergence speed and estimation accuracy, particularly in minimizing the sum of squared errors. The proposed framework proves to be a robust and versatile tool for real-world parameter estimation in fuel cells, providing a reliable computational foundation for efficiency-oriented PEMFC modeling and analysis.