Niu, Tong and Zhang, Caizhi and Ren, Yong and Yang, Lei and Wang, Zihao and Rodić, Aleksandar D. and Jeločnik, Marko and Simonović, Miloš (2025) A fast online prediction method for the health state and electrochemical performance of proton exchange membrane fuel cells without prior modeling. Journal of Power Sources, 630. ISSN 1873-2755
Full text not available from this repository.Abstract
In fuel cell technology, Proton Exchange Membrane Fuel Cells (PEMFC) require rapid online prediction of their electrochemical performance to extend their service life. However, traditional methods exhibit limitations in real-time capturing of PEMFC performance variations and predicting future states. This study introduces an innovative method that combines the distribution of relaxation times (DRT) analysis, linear regression, and neural networks. Firstly, historical electrochemical impedance spectroscopy (EIS) data is used to obtain DRT spectra. By integrating the peak values in the DRT spectrum curve, key resistance parameters in the equivalent circuit are obtained. Subsequently, a linear regression model is established using online stack voltage and impedance data. Finally, the neural network model predicts the future voltage and combines it with a linear regression model to obtain the impedance spectrum of the PEMFC at future times. The results show that at specific times and frequencies, the fitting accuracy of key feature parameters reaches 98.8529 % and 93.1956 %, respectively, with a calculation error of 6.76948 × 10^(-6), verifying the effectiveness of the method. Therefore, acquiring partial EIS data during the initial stage of stack operation, in conjunction with the proposed method, suffices for achieving high-precision online prediction of PEMFC electrochemical performance.
| Item Type: | Article |
|---|---|
| Uncontrolled Keywords: | Distribution of relaxation times, electrochemical impedance spectroscopy, linear regression model state of health |
| Depositing User: | Unnamed user with email srdjan.jurlina@ien.bg.ac.rs |
| Date Deposited: | 06 May 2025 10:48 |
| Last Modified: | 06 May 2025 10:48 |
| URI: | http://repository.iep.bg.ac.rs/id/eprint/1080 |
Actions (login required)
 |
View Item |
