| INORGANIC MATERIALS AND CERAMIC MATRIX COMPOSITES |
|
|
|
|
|
| Study on Performance Degradation Prediction of Proton Exchange Membrane Fuel Cells Combining Mechanistic Information and Deep Learning Algorithms |
| YU Wanqiu1,2, WU Fan1,2, JIANG Panxing1,2, WANG Linhui1,2, WANG Yadong1,2, ZHAN Zhigang1,2,*
|
1 State Key Laboratory of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology, Wuhan 430070, China
2 Hubei Key Laboratory of Fuel Cells, Wuhan University of Technology, Wuhan 430070, China |
|
|
|
|
Abstract Proton exchange membrane fuel cell (PEMFC) is an important modern sustainable energy conversion device. Accurate prediction of its performance degradation trend is of great significance for improving system reliability and prolonging service life. The conventional data-driven method lacks the consideration of mechanism information, which leads to the ‘black box’ phenomenon of neural network model. This paper proposes a hybrid deep learning model CNN-LSTM-A, which integrates convolutional neural network (CNN), long short-term memory network (LSTM) and attention mechanism, and introduces sparrow search algorithm (SSA) to realize automatic optimization of hyperparameters. Taking the output voltage of PEMFC as the degradation performance index, the performance of the model is evaluated under different training set partition ratios. The results show that the model can accurately capture the aging trend of fuel cells. In order to further enhance the physical interpretability of the model, this paper introduces the mechanism parameters related to the fuel cell degradation mechanism as input features. The results show that the prediction accuracy of the model is improved after the fusion of mechanism information, which verifies the importance of mechanism information to the neural network model in the prediction task. In addition, the superiority of the model proposed in this paper is verified by comparative experiments. The results show that the introduction of attention mechanism and SSA significantly improves the performance of the model, and the fusion model of CNN and LSTM has better prediction effect than the single structure.
|
|
Published:
Online: 2026-04-16
|
|
|
|
|
1 Jiao K, Xuan J, Du Q, et al. Nature, 2021, 595(7867), 361.
2 Zhang C, Zhang Y, Wang L, et al. Renewable and Sustainable Energy Reviews, 2023, 182, 113369.
3 Petrone G, Zamboni W, Spagnuolo G. Applied Energy, 2019, 242, 1226.
4 Zhu D, Yang B, Liu Y, et al. Applied Energy, 2022, 311, 118636.
5 Zhu D, Yang B, Liu Q, et al. Applied Energy, 2020, 272, 115225.
6 Hua Z, Zheng Z, Pahon E, et al. Journal of Power Sources, 2022, 529, 231256.
7 Liu H, Chen J, Hissel D, et al. Renewable and Sustainable Energy Reviews, 2020, 123, 109721.
8 Chen H, Zhan Z, Jiang P, et al. Applied Energy, 2022, 310, 118556.
9 Zhang X, Yang D, Luo M, et al. International Journal of Hydrogen Energy, 2017, 42(16), 11868.
10 Lu L, Ouyang M, Huang H, et al. Journal of Power Sources, 2007, 164(1), 306.
11 Ou M, Zhang R, Shao Z, et al. Journal of Power Sources, 2021, 488, 229435.
12 Jouin M, Gouriveau R, Hissel D, et al. Reliability Engineering & System Safety, 2016, 148, 78.
13 Ma R, Yang T, Breaz E, et al. Applied Energy, 2018, 231, 102.
14 Morando S, Jemei S, Hissel D, et al. International Journal of Hydrogen Energy, 2017, 42(2), 1472.
15 Javed K, Gouriveau R, Zerhouni N, et al. Journal of Power Sources, 2016, 324, 745.
16 Tu C, Zhou F, Pan M. International Journal of Hydrogen Energy, 2024, 74, 414.
17 Yang J, Wang L, Zhang B, et al. Energy, 2024, 291, 130334.
18 Yi F, Shu X, Zhou J, et al. International Journal of Hydrogen Energy, 2025, 111, 228.
19 Liu J, Li Q, Chen W, et al. International Journal of Hydrogen Energy, 2019, 44(11), 5470.
20 Jia C, He H, Zhou J, et al. International Journal of Hydrogen Energy, 2024, 60, 133.
21 Li S, Luan W, Wang C, et al. International Journal of Hydrogen Energy, 2022, 47(78), 33466.
22 Fukushima K. Biological Cybernetics, 1980, 36(4), 193.
23 Li J, Bastani C. AIP Advances, 2025, 15(2), 025117.
24 Hochreiter S, Schmidhuber J. Neural Computation, 1997, 9(8), 1735.
25 Xue J, Shen B. Systems Science & Control Engineering, 2020, 8(1), 22. |
|
|
|
渝公网安备50019002502923号 © Editorial Office of Materials Reports.
2026, Vol. 40 
