Roles of Activation Functions on the Polarization Modeling for the 1.2 kWp Proton Exchange  Membrane Fuel Cells: A Focus on ReLU, Sigmoid, and Tanh

Chanin  Khamwandee; Arnusorn Saengprajak

doi:10.69650/rast.2026.263264

Authors

Chanin Khamwandee Energy Research Unit, Department of Physics, Faculty of Science, Mahasarakham University, Maha Sarakham 44150, Thailand
Arnusorn Saengprajak Energy Research Unit, Department of Physics, Faculty of Science, Mahasarakham University, Maha Sarakham 44150, Thailand

DOI:

https://doi.org/10.69650/rast.2026.263264

Keywords:

Fuel Cell, Modelling , ANN, Activation Function, Polarization Curve, ANNOVA, LabVIEW

Abstract

Artificial neural network (ANN) analysis techniques have gained increasing popularity across diverse engineering applications, particularly in the multi-physics modeling of fuel cell systems. However, previous literature lacks a comprehensive comparison of the effectiveness of fundamental nonlinear activation functions in ANN architectures specifically tailored for the mid-sized proton exchange membrane fuel cell (PEMFC). This study investigates the influence of three primary activation functions—the rectified linear unit, logistic sigmoid, and hyperbolic tangent—on the predictive modeling of a 1.2 kWp PEMFC. The ANN model was developed within LabVIEW to simulate voltage across a dynamic operation, including variations in temperature, reactant pressure, and load current. The simulation outputs for each activation function were rigorously validated against empirical data using analysis of variance and comprehensive error analysis, including root mean square error. The results indicate that while the rectified linear unit and hyperbolic tangent models achieved performance within statistically acceptable margins, the logistic sigmoid function demonstrated superior accuracy in capturing the nonlinear voltage–current relationship. Specifically, the sigmoid function provided a more nuanced representation of the internal loss—namely, activation and concentration polarizations—characteristics associated with this scale of PEMFC systems. This study reveals that the superior sensitivity of the sigmoid function makes it the most suitable choice for digital twin diagnostics and real-time control of mid-sized PEMFC stacks.

References

Kumar, L. A., Renuka, D. K., Rose, S. L., Shunmuga priya, M. C. and Wartana, I. M., Deep learning based assistive technology on audio visual speech recognition for hearing impaired. International Journal of Cognitive Computing in Engineering. 3 (2022) 24-30, doi: https://doi.org/10.1016/j.ijcce.2022.01.003.

Jackulin, C. and Murugavalli, S., A comprehensive review on detection of plant disease using machine learning and deep learning approaches. Measurement: Sensors. 24 (2022) 100441, doi: https://doi.org/10.1016/j.measen.2022.100441.

Bouraya, S. and Belangour, A., A comparative analysis of activation functions in neural networks: unveiling categories. Bulletin of Electrical Engineering and Informatics. 13 (2024) 3301-3308, doi: https://doi.org/10.11591/eei.v13i5.7274.

Le, T. H., Thakur, D. and Nguyen, P. K. T., Modeling and optimization of direct urea-hydrogen peroxide fuel cell using the integration of artificial neural network and bio-inspired algorithms. Journal of Electroanalytical Chemistry. 922 (2022) 116783, doi: https://doi.org/10.1016/j.jelechem.2022.116783.

Tian, P. Performance prediction of PEM fuel cell using artificial neural network machine learning. Report No. 28022774, University of California, Irvine, (2020) 1-52.

Subotić, V., Eibl, M. and Hochenauer, C., Artificial intelligence for time-efficient prediction and optimization of solid oxide fuel cell performances. Energy Conversion and Management. 230 (2021) 113764, doi: https://doi.org/10.1016/j.enconman.2020.113764.

Legala, A., Zhao, J. and Li, X., Machine learning modeling for proton exchange membrane fuel cell performance. Energy and AI. 10 (2022) 100183, doi: https://doi.org/10.1016/j.egyai.2022.100183.

Obermaier, M., Rauber, M., Bauer, A., Lochner, T., Du, F. and Scheu, C., Local Fuel Starvation Degradation of an Automotive PEMFC Full Size Stack. Fuel Cells. 20 (2020) 394-402, doi: https://doi.org/10.1002/fuce.201900180.

Al-Baghdadi, M., Three-dimensional solid mechanics-CFD modeling of a PEM fuel cell stack. International Journal of Energy and Environment. 9 (2018) 1-26.

Mastropasqua, L., Campanari, S. and Brouwer, J., Solid oxide fuel cell short stack performance testing - Part A: Experimental analysis and μ-combined heat and power unit comparison. Journal of Power Sources. 371 (2017) 225-237, doi: https://doi.org/10.1016/j.jpowsour.2017.10.028.

Zhou, Y., Lin, G., Shih, A. J. and Hu, S. J., Multiphysics Modeling of Assembly Pressure Effects on Proton Exchange Membrane Fuel Cell Performance. Journal of Fuel Cell Science and Technology. 6 (2009) 1-7, doi: https://doi.org/10.1115/1.3081426.

Tochigi, S. and Dowaki, K., Environmental Impact Assessment of PEM Fuel Cell Combined Heat and Power Generation System for Residential Application Considering Cathode Catalyst Layer Degradation. Energies. 16 (2023) 1985, doi: https://doi.org/10.3390/en16041985.

Islam, M. S., Shudo, Y. and Hayami, S., Energy Conversion and Storage in Fuel Cells and Super-Capacitors from Chemical Modifications of Carbon Allotropes: State-of-Art and Prospect. Bulletin of the Chemical Society of Japan. 95 (2021) 1-25, doi: https://doi.org/10.1246/bcsj.20210297.

Legala, A., Shahgaldi, S. and Li, X., Data-based modelling of proton exchange membrane fuel cell performance and degradation dynamics. Energy Conversion and Management. 296 (2023) 117668, doi: https://doi.org/10.1016/j.enconman.2023.117668.

Pei, P., Fu, X., Zhu, Z., Ren, P. and Chen, D., Activation of polymer electrolyte membrane fuel cells: Mechanisms, procedures, and evaluation. International Journal of Hydrogen Energy. 47 (2022) 24897-24915, doi: https://doi.org/10.1016/j.ijhydene.2022.05.228.

Du, F., Dao, T. A., Peitl, P. V. J., Bauer, A., Preuss, K., Bonastre, A. M., Sharman, J., Spikes, G., Perchthaler, M., Schmidt, T. J. and Orfanidi, A., Effects of PEMFC Operational History under Dry/Wet Conditions on Additional Voltage Losses due to Ionomer Migration. Journal of The Electrochemical Society. 167 (2020) 144513, doi: https://doi.org/10.1149/1945-7111/abc83f.

Kumar, R. and Subramanian, K. A., Analysis of activation, ohmic, and concentration losses in hydrogen fuelled PEM fuel cell. Advances in Energy Research. 8 (2022) 253–264, doi: https://doi.org/10.12989/eri.2022.8.4.253.

Yang, B., Liang, B., Qian, Y., Zheng, R., Su, S., Guo, Z. and Jiang, L., Parameter identification of PEMFC via feedforward neural network-pelican optimization algorithm. Applied Energy. 361 (2024) 122857, doi: https://doi.org/10.1016/j.apenergy.2024.122857.

Alsaeedi, A. H., Aljanabi, A. H., Manna, M. E. and Albukhnefis, A. L., A proactive metaheuristic model for optimizing weights of artificial neural network. Indonesian Journal of Electrical Engineering and Computer Science. 20 (2020) 976-984, doi: http://doi.org/10.11591/ijeecs.v20.i2.pp976-984.

Rasamoelina, A. D., Adjailia, F. and Sinčák, P. A Review of Activation Function for Artificial Neural Network. in 2020 IEEE 18th World Symposium on Applied Machine Intelligence and Informatics (SAMI). (2020), 281-286, doi: https://doi.org/10.1109/SAMI48414.2020.9108717.

Sharma, S., Sharma, S. and Athaiya, A., Activation functions in neural networks. International Journal of Engineering Applied Sciences and Technology. 4 (2020) 310–316, doi: https://doi.org/10.33564/IJEAST.2020.V04I12.054.

Nanadegani, F. S., Lay, E. N., Iranzo, A., Salva, J. A. and Sunden, B., On neural network modeling to maximize the power output of PEMFCs. Electrochimica Acta. 348 (2020) 136345, doi: https://doi.org/10.1016/j.electacta.2020.136345.

Zuo, B., Zhang, Z., Cheng, J., Huo, W., Zhong, Z. and Wang, M., Data-driven flooding fault diagnosis method for proton-exchange membrane fuel cells using deep learning technologies. Energy Conversion and Management. 251 (2022) 115004, doi: https://doi.org/10.1016/j.enconman.2021.115004.

Deng, S., Zhang, J., Zhang, C., Luo, M., Ni, M., Li, Y. and Zeng, T., Prediction and optimization of gas distribution quality for high-temperature PEMFC based on data-driven surrogate model. Applied Energy. 327 (2022) 120000, doi: https://doi.org/10.1016/j.apenergy.2022.120000.

Bayrak, Z. U. and Celik, N., Determining the Effects of Operating Conditions on Current Density of a PEMFC by Using Taguchi Method and ANOVA. Arabian Journal for Science and Engineering. 49 (2024) 10741-10752, doi: https://doi.org/10.1007/s13369-023-08470-9.

Ariesanti, Y., Octavianus, P. G. M., Handayani, A. T. and Abbas, B., Characterization of Polyvinyl Alcohol–Collagen–Hydroxyapatite Composite Membrane from Lates calcarifer Scales for Guided Tissue and Bone Regeneration. European Journal of Dentistry. 17 (2023) 1153-1162, doi: https://doi.org/10.1055/s-0042-1759488.

Sadeghi, H., Mirzaee, I., Khalilarya, S. and Ahmadi, N., Numerical investigation of gas channel geometry of proton exchange membrane fuel cell. Renewable Energy Research and Applications. 1 (2020) 93-114, doi: https://doi.org/10.22044/rera.2020.9182.1021.

Mota, A., Gonzalez, E. R. and Eiswirth, M., Kinetic insights over a PEMFC operating on stationary and oscillatory states. The Journal of Physical Chemistry A. 115 (2011) 13773-13782, doi: https://doi.org/10.1021/jp205341w.

Puranik, S. V., Keyhani, A. and Khorrami, F. Neural Network Modeling of Proton Exchange Membrane Fuel Cell. IEEE Transactions on Energy Conversion. 25 (2010) 474-483, doi: https://doi.org/10.1109/TEC.2009.2035691.

Doumbia, M. L., PEM fuel cell modelling using artificial neural networks. International Journal of Renewable Energy Research. 4 (2014) 725-730.

Lau, M. M. and Lim, K. H. Review of Adaptive Activation Function in Deep Neural Network. in 2018 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES). (2018), 686-690, doi: https://doi.org/10.1109/IECBES.2018.8626714.

Wang, X., Qin, Y., Wang, Y., Xiang, S., and Chen, H., ReLTanh: An activation function with vanishing gradient resistance for SAE-based DNNs and its application to rotating machinery fault diagnosis. Neurocomputing. 363 (2019), 88-98, doi: https://doi.org/10.1016/j.neucom.2019.07.017.