Hydrogen fuel cell parameter estimation using an innovative hybrid estimation model based on deep learning and probability pooling
Yazarlar (2)
Prof. Dr. Cem EMEKSİZ Tokat Gaziosmanpaşa Üniversitesi, Türkiye
Prof. Dr. Mustafa Tan Tokat Gaziosmanpaşa Üniversitesi, Türkiye
| Makale Türü | Özgün Makale (SSCI, AHCI, SCI, SCI-Exp dergilerinde yayınlanan tam makale) | ||
| Dergi Adı | International Journal of Hydrogen Energy (Q1) | ||
| Dergi ISSN | 0360-3199 Wos Dergi Scopus Dergi | ||
| Dergi Tarandığı Indeksler | SCI-Expanded | ||
| Makale Dili | İngilizce | Basım Tarihi | 03-2024 |
| Cilt / Sayı / Sayfa | 110 / 1 / 445–456 | DOI | 10.1016/j.ijhydene.2025.02.272 |
| Makale Linki | https://www.sciencedirect.com/science/article/pii/S0360319925008493 | ||
| UAK Araştırma Alanları |
Yenilenebilir Enerji Sistemleri
|
||
| Özet |
| Developing a suitable model for the reliable performance of Polymer Electrolyte Membrane Fuel Cells (PEMFC) is important. However, the nonlinearity, computational complexity of the mathematical models used and the electrochemical nature of the system make it difficult to model. One of these challenges is the determination of model parameters. Parameters are usually determined using experimental data or numerical simulations, but this process is time-consuming and costly. Therefore, the Probability pool method and Convolutional Neural Network were used in this study for the first time to determine the PEMFC parameters. This hybrid approach constitutes the unique and innovative aspect of the study. With the model created, a classification map was used to identify the PEMFC parameters more accurately and quickly. This study offers a new perspective on data-driven model structures in order to provide solutions against limited data and random factors. The results show that the performance metrics exhibit high accuracy (e.g., R2 = 0.9997, RSME = 0.0066 and MAPE = 0.0068 for Voltage1). This approach aimed to reduce the need for expensive and time-consuming experimental studies and to enable the development of more efficient PEMFCs for clean energy applications. |
| Anahtar Kelimeler |
| Convolutional neural network | Deep learning | Hydrogen production | Parameter estimation | Polymer electrolyte membrane fuel cells |
Science Direct
Hydrogen fuel cell parameter estimation using an innovative hybrid estimation model based on deep learning and probability pooling
BM Sürdürülebilir Kalkınma Amaçları
| Atıf Sayıları | |
| Web of Science | 2 |
| Scopus | 2 |