Journal of Zhejiang University

Journal of Zhejiang University SCIENCE A

Accepted manuscript available online (unedited version)

Real-time degradation modeling for automotive PEMFC stacks: a multi-scale fusion network validated on an industrial 215-channel system

Author(s): Zifei WANG, Xiangxian ZHU, Congxin LI, Daidai CHEN, Zhitao LIU, Longhua MA, Jili TAO, Hongye SU
Affiliation(s): School of Information Science and Engineering, NingboTech University, Ningbo 315100, China; more
Corresponding email(s): taojili@nbt.edu.cn
Key Words: Large-scale proton exchange membrane fuel cell (PEMFC) stacks, Multi-scale bidirectional fusion network (MBFNet), Degradation prediction, Automotive scenarios

<<< Previous Paper \|Next Paper >>>

Zifei WANG, Xiangxian ZHU, Congxin LI, Daidai CHEN, Zhitao LIU, Longhua MA, Jili TAO, Hongye SU. Real-time degradation modeling for automotive PEMFC stacks: a multi-scale fusion network validated on an industrial 215-channel system[J]. Journal of Zhejiang University Science A,in press.Frontiers of Information Technology & Electronic Engineering,in press.https://doi.org/10.1631/jzus.A2500337

@article{title="Real-time degradation modeling for automotive PEMFC stacks: a multi-scale fusion network validated on an industrial 215-channel system",
author="Zifei WANG, Xiangxian ZHU, Congxin LI, Daidai CHEN, Zhitao LIU, Longhua MA, Jili TAO, Hongye SU",
journal="Journal of Zhejiang University Science A",
year="in press",
publisher="Zhejiang University Press & Springer",
doi="https://doi.org/10.1631/jzus.A2500337"
}

%0 Journal Article
%T Real-time degradation modeling for automotive PEMFC stacks: a multi-scale fusion network validated on an industrial 215-channel system
%A Zifei WANG
%A Xiangxian ZHU
%A Congxin LI
%A Daidai CHEN
%A Zhitao LIU
%A Longhua MA
%A Jili TAO
%A Hongye SU
%J Journal of Zhejiang University SCIENCE A
%P 506-517
%@ 1673-565X
%D in press
%I Zhejiang University Press & Springer
doi="https://doi.org/10.1631/jzus.A2500337"

TY - JOUR
T1 - Real-time degradation modeling for automotive PEMFC stacks: a multi-scale fusion network validated on an industrial 215-channel system
A1 - Zifei WANG
A1 - Xiangxian ZHU
A1 - Congxin LI
A1 - Daidai CHEN
A1 - Zhitao LIU
A1 - Longhua MA
A1 - Jili TAO
A1 - Hongye SU
J0 - Journal of Zhejiang University Science A
SP - 506
EP - 517
%@ 1673-565X
Y1 - in press
PB - Zhejiang University Press & Springer
ER -
doi="https://doi.org/10.1631/jzus.A2500337"

Abstract
Chinese Summary
Academic Network
Reviewer Comment

Abstract: Accurately predicting long-term degradation patterns in proton exchange membrane fuel cell (PEMFC) stacks under automotive operating conditions remains challenging. Prediction methods are largely constrained by laboratory-scale experiments and limited stack sizes, resulting in insufficient accuracy and generalization capability. To address these limitations, in this paper we propose a multi-scale bidirectional fusion network (MBFNet) tailored for an industrial 215-channel PEMFC stack, enabling accurate degradation prediction under accelerated real-world dynamic conditions using gas-heat-electricity (GHE) co-simulation data. A channel-joint adaptive noise correlation threshold (NCT) algorithm is introduced to account for variable correlations across sensors and operating conditions without relying on prior physical modeling. A multi-scale decomposition module captures degradation dynamics at different temporal scales, while a bidirectional fusion module integrates global trends and local details into the final prediction. Experimental results show that MBFNet achieves 18.6% lower prediction error and 36.8% fewer parameters than the long short-term memory (LSTM)-attention benchmark under real operating scenarios. In multi-step prediction tasks, MBFNet reduces root mean square error by an average of 24.5% relative to LSTM-attention and 55.2% relative to a one-dimensional convolutional neural network (1D-CNN) across four prediction horizons, better satisfying automotive application requirements. Moreover, MBFNet exhibits strong physical interpretability, making it efficient to implement and promising for practical deployment.

汽车质子交换膜燃料电池堆实时退化建模：基于工业级215通道系统验证的多尺度融合网络

作者：王子非^1,2，朱想先³，李从心⁴，陈岱岱³，刘之涛²，马龙华¹，陶吉利¹，苏宏业²
机构：¹浙大宁波理工学院，信息科学与工程学院，中国宁波，315100；²浙江大学，工业控制技术全国重点实验室，中国杭州，310027；³宁波均胜电子股份有限公司，中国宁波，315040；⁴宁波绿动氢能科技研究院有限公司，中国宁波，315033
目的：准确预测工业级215通道车用质子交换膜燃料电池电堆在动态工况下的长期退化行为，解决传统方法在实验室条件、小规模电堆和噪声干扰下的预测精度低、泛化能力差的问题，为燃料电池汽车的健康管理与寿命预测提供可靠模型。
创新点：1.提出多尺度双向融合网络，融合通道联合自适应噪声相关阈值去噪算法，实现无先验建模的多物理场噪声抑制；2.引入多尺度分解模块解耦电压恢复与老化趋势；3.设计轻量化双向融合模块，在提升预测精度的同时显著减少参数量，适配车载边缘计算资源限制。
方法：1.构建215通道联合仿真平台采集实车工况退化数据；2.采用噪声相关阈值算法动态抑制传感器噪声；3.通过多尺度分解提取不同时间尺度的退化特征；4.利用双向融合模块整合电堆级趋势与单电池级波动；5.基于均方根误差、平均绝对误差等指标与长短期记忆（LSTM）、门控循环单元等模型进行单步/多步预测对比及消融实验验证。
结论：1.多尺度双向融合网络在关键退化阶段（285.0→255.0 V）实现了0.0180的均方根误差，比LSTM-attention提升18.6%；2.多步预测误差较LSTM-attention和一维卷积神经网络分别降低24.5%与55.2%；3.模型参数量减少36.8%，满足车载电子控制单元资源限制，具备良好的工程应用前景与物理可解释性。

关键词组：大型质子交换膜燃料电池堆；多尺度双向融合网络；退化预测；汽车场景

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference

[1]BenaggouneK, YueML, JemeiS, et al., 2022. A data-driven method for multi-step-ahead prediction and long-term prognostics of proton exchange membrane fuel cell. Applied Energy, 313:118835.

[2]BresselM, HilairetM, HisselD, et al., 2016. Extended Kalman filter for prognostic of proton exchange membrane fuel cell. Applied Energy, 164:220-227.

[3]ChenK, LaghroucheS, DjerdirA, 2019. Degradation model of proton exchange membrane fuel cell based on a novel hybrid method. Applied Energy, 252:113439.

[4]ChenK, BadjiA, LaghroucheS, et al., 2022. Polymer electrolyte membrane fuel cells degradation prediction using multi-kernel relevance vector regression and whale optimization algorithm. Applied Energy, 318:119099.

[5]ChengYJ, ZerhouniN, LuC, 2018. A hybrid remaining useful life prognostic method for proton exchange membrane fuel cell. International Journal of Hydrogen Energy, 43(27):12314-12327.

[6]HeK, LiuZY, SunYN, et al., 2022. Degradation prediction of proton exchange membrane fuel cell using auto-encoder based health indicator and long short-term memory network. International Journal of Hydrogen Energy, 47(82):35055-35067.

[7]HeWB, LiuT, MingWY, et al., 2024. Progress in prediction of remaining useful life of hydrogen fuel cells based on deep learning. Renewable and Sustainable Energy Reviews, 192:114193.

[8]JouinM, GouriveauR, HisselD, et al., 2014. Prognostics of PEM fuel cell in a particle filtering framework. International Journal of Hydrogen Energy, 39(1):481-494.

[9]LiCZ, LinW, WuHY, et al., 2023. Performance degradation decomposition-ensemble prediction of PEMFC using CEEMDAN and dual data-driven model. Renewable Energy, 215:118913.

[10]LiuH, ChenJ, HisselD, et al., 2019. A multi-scale hybrid degradation index for proton exchange membrane fuel cells. Journal of Power Sources, 437:226916.

[11]LiuMY, WangC, ZhangJB, et al., 2014. Diagnosis of membrane electrode assembly degradation with drive cycle test technique. International Journal of Hydrogen Energy, 39(26):14370-14375.

[12]LiuZY, MaoL, HuZY, et al., 2022. A novel densely connected neural network for proton exchange membrane fuel cell fault diagnosis. International Journal of Hydrogen Energy, 47(94):40041-40053.

[13]MaR, YangT, BreazE, et al., 2018. Data-driven proton exchange membrane fuel cell degradation predication through deep learning method. Applied Energy, 231:102-115.

[14]MaR, ChaiXY, GengRX, et al., 2023. Recent progress and challenges of multi-stack fuel cell systems: fault detection and reconfiguration, energy management strategies, and applications. Energy Conversion and Management, 285:117015.

[15]RenP, PeiPC, LiYH, et al., 2020. Degradation mechanisms of proton exchange membrane fuel cell under typical automotive operating conditions. Progress in Energy and Combustion Science, 80:100859.

[16]SahajpalK, RanaKPS, KumarV, 2023. Accurate long-term prognostics of proton exchange membrane fuel cells using recurrent and convolutional neural networks. International Journal of Hydrogen Energy, 48(78):30532-30555.

[17]ShinD, YooS, 2023. Diagnostic method for PEM fuel cell states using probability distribution-based loss component analysis for voltage loss decomposition. Applied Energy, 330:120340.

[18]SunW, YanR, JinR, et al., 2024. LiteFormer: a lightweight and efficient Transformer for rotating machine fault diagnosis. IEEE Transactions on Reliability, 73(2):1258-1269.

[19]TianL, GaoY, YangHY, et al., 2025. Multi-scenario long-term degradation prediction of PEMFC based on generative inference informer model. Applied Energy, 377:124398.

[20]TianZ, WangJH, Al-DurraA, et al., 2023. A novel aging prediction method of fuel cell based on empirical mode decomposition and complexity threshold quantitative criterion. Journal of Power Sources, 574:233120.

[21]WangC, DouMF, LiZL, et al., 2023. Data-driven prognostics based on time-frequency analysis and symbolic recurrent neural network for fuel cells under dynamic load. Reliability Engineering System Safety, 233:109123.

[22]WangPH, LiuH, ChenJ, et al., 2021. A novel degradation model of proton exchange membrane fuel cells for state of health estimation and prognostics. International Journal of Hydrogen Energy, 46(61):31353-31361.

[23]WilberforceT, OlabiAG, 2021. Proton exchange membrane fuel cell performance prediction using artificial neural network. International Journal of Hydrogen Energy, 46(8):6037-6050.

[24]WilberforceT, AlaswadA, Garcia–PerezA, et al., 2023. Remaining useful life prediction for proton exchange membrane fuel cells using combined convolutional neural network and recurrent neural network. International Journal of Hydrogen Energy, 48(1):291-303.

[25]WuYM, BreazE, GaoF, et al., 2016. A modified relevance vector machine for PEM fuel-cell stack aging prediction. IEEE Transactions on Industry Applications, 52(3):2573-2581.

[26]XiaZT, WangYN, MaLH, et al., 2022. A hybrid prognostic method for proton-exchange-membrane fuel cell with decomposition forecasting framework based on AEKF and LSTM. Sensors, 23(1):166.

[27]YiFY, ShuX, ZhouJM, et al., 2025. Remaining useful life prediction of PEMFC based on matrix long short-term memory. International Journal of Hydrogen Energy, 111:228-237.

[28]ZhouDM, GaoF, BreazE, et al., 2017. Degradation prediction of PEM fuel cell using a moving window based hybrid prognostic approach. Energy, 138:1175-1186.

[29]ZhouSW, ShearingPR, BrettDJL, et al., 2022. Machine learning as an online diagnostic tool for proton exchange membrane fuel cells. Current Opinion in Electrochemistry, 31:100867.

[30]ZuoJ, LvH, ZhouDM, et al., 2021. Deep learning based prognostic framework towards proton exchange membrane fuel cell for automotive application. Applied Energy, 281:115937.

Open peer comments: Debate/Discuss/Question/Opinion

<1>