Transient modeling of a solid oxide fuel cell using an efficient deep learning HY-CNN-NARX paradigm
Source abstract
Abstract
Control and monitoring systems are crucial for ensuring optimal performance, efficiency, and longevity of Solid Oxide Fuel Cells (SOFC). Developing a model to accurately capture the transient behavior of the SOFC under dynamic operation is essential for these applications. As electrochemical, mass-transfer, and thermal equations are involved in SOFCs’ dynamic, physics-based approaches to capture all the complexity of SOFC systems are often too computationally expensive for real-time implementation. In this paper, an innovative multi-input neural network is developed to build an accurate model of SOFC that is computationally efficient. The proposed algorithm uses experimental data and is a modified nonlinear autoregressive exogenous (NARX) network. An optimal sequence of the most recent observations (output voltages) that characterize the SOFC dynamics is passed through stacked 1D convolutional layers (CNN) to extract latent spatial/temporal information. Then the output is concatenated with current and past exogenous inputs. The fusion of learned features, representing the internal dynamics and the effect of exogenous inputs, is then fed to a fully connected network for one-step ahead performance prediction. This hybrid structure, called HY-CNN-NARX, is capable of identifying the transient dynamics of the SOFCs by unrolling information from historical outputs and inputs within a feedforward framework. To evaluate the model performance, lab-scale tubular SOFCs are experimentally tested at 650–750 °C under various dynamic operations and initial conditions, and the transient and steady-state responses of the SOFC are collected. Once the model is validated using a wide operating range of data, the generalization and prediction performance of the proposed network is compared with a conventional NARX and a recurrent stacked Long short-term memory (LSTM) model. The comparative results demonstrate that the proposed HY-CNN-NARX model outperforms the NARX model on unseen data with an accuracy improvement of 59.2% for root mean square error and 53.7% for mean absolute percentage error. In addition, the prediction performance of the HY-CNN-NARX is comparable to its recurrent counterpart, the LSTM model, but with a 39.3% faster execution time.
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Fuel Cells and Related Materials · Advancements in Solid Oxide Fuel Cells · Gas Sensing Nanomaterials and Sensors
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
References (62)
Show source-provided references
1. Progress and outlook for solid oxide fuel cells for transportation applications10.1038/s41929-019-0310-y ↗
2. A comprehensive review of solid oxide fuel cells operating on various promising alternative fuels10.1016/j.enconman.2021.115175 ↗
3. Investigating the performance of a solid oxide electrolyzer multi-stack module with a multiphysics homogenized model10.1016/j.jpowsour.2023.234019 ↗
4. A systematic review of machine learning methods applied to fuel cells in performance evaluation, durability prediction, and application monitoring10.1016/j.ijhydene.2022.10.261 ↗
5. A control oriented low order dynamic model for planar SOFC using minimum gibbs free energy method10.1016/j.jpowsour.2006.12.009 ↗
6. Solid oxide fuel cell modeling10.1109/TIE.2008.2009516 ↗
7. A versatile computational tool for model-based design, control and diagnosis of a generic solid oxide fuel cell integrated stack module10.1016/j.enconman.2018.06.062 ↗
8. Control-oriented modelling and analysis of a solid oxide fuel cell system10.1016/j.ijhydene.2020.02.061 ↗
9. Modeling of solid oxide fuel cell (SOFC) electrodes from fabrication to operation: correlations between microstructures and electrochemical performances10.1016/j.enconman.2019.04.002 ↗
10. Multiphysics transients modeling of solid oxide fuel cells: methodology of circuit equivalents and use in EMTP-type power system simulation10.1109/TEC.2017.2687886 ↗
11. Modeling of anode-supported SOFCs with samaria doped-ceria electrolytes operating at 500–600 °C10.1016/j.jpowsour.2010.01.013 ↗
12. Three-dimensional multi-physics modelling and structural optimization of SOFC large-scale stack and stack tower10.1016/j.ijhydene.2022.10.146 ↗
13. 3D modeling of anode-supported planar SOFC with internal reforming of methane10.1016/j.jpowsour.2007.07.003 ↗
14. Multidimensional reversible solid oxide fuel cell modeling for embedded applications10.1109/TEC.2017.2762962 ↗
15. Solid oxide fuel cell (SOFC) performance evaluation, fault diagnosis and health control: A review10.1016/j.jpowsour.2021.230058 ↗
16. Artificial intelligence for solid oxide fuel cells: Combining automated high accuracy artificial neural network model generation and genetic algorithm for time-efficient performance prediction and optimization10.1016/j.enconman.2023.117263 ↗
17. Constrained model predictive control of a solid oxide fuel cell based on genetic optimization10.1016/j.jpowsour.2011.03.010 ↗
18. A neural network estimator of solid oxide fuel cell performance for on-field diagnostics and prognostics applications10.1016/j.jpowsour.2013.04.114 ↗
19. Artificial neural network model of a short stack solid oxide fuel cell based on experimental data10.1016/j.jpowsour.2013.08.018 ↗
20. Comparison of support vector regression and random forest algorithms for estimating the SOFC output voltage by considering hydrogen flow rates10.1016/j.ijhydene.2020.07.265 ↗
21. Modeling the SOFC by BP neural network algorithm10.1016/j.ijhydene.2021.03.132 ↗
22. Optimizing the homogeneity and efficiency of a solid oxide electrolysis cell based on multiphysics simulation and data-driven surrogate model10.1016/j.jpowsour.2023.232760 ↗
23. Performance prediction model of solid oxide fuel cell system based on neural network autoregressive with external input method10.3390/pr8070828 ↗
24. Model identification and strategy application for solid oxide fuel cell using rotor hopfield neural network based on a novel optimization method10.1016/j.ijhydene.2020.07.127 ↗
25. An optimal model identification for solid oxide fuel cell based on extreme learning machines optimized by improved red fox optimization algorithm10.1016/j.ijhydene.2021.06.046 ↗
26. Model identification of solid oxide fuel cell using hybrid elman neural network/quantum pathfinder algorithm10.1016/j.egyr.2021.05.070 ↗
27. Emulating spatial and temporal outputs from fuel cell and battery models: A comparison of deep learning and Gaussian process models10.1115/1.4054195 ↗
28. Generalized spatial-temporal fault location method for solid oxide fuel cells using LSTM and causal inference10.1109/TTE.2022.3187870 ↗
29. Data-driven voltage prognostic for solid oxide fuel cell system based on deep learning10.3390/en15176294 ↗
30. Data-driven approaches for predicting performance degradation of solid oxide fuel cells system considering prolonged operation and shutdown accumulation effect10.1016/j.jpowsour.2024.234186 ↗
31. Application of ELM–Hammerstein model to the identification of solid oxide fuel cells10.1007/s00521-016-2453-y ↗
32. A novel multi-physics and multi-dimensional model for solid oxide fuel cell stacks based on alternative mapping of BP neural networks10.1016/j.jpowsour.2021.229784 ↗
33. Coupling deep learning and multi-objective genetic algorithms to achieve high performance and durability of direct internal reforming solid oxide fuel cell10.1016/j.apenergy.2022.119046 ↗
34. Correlation between microstructures and macroscopic properties of nickel/yttria-stabilized zirconia (Ni-YSZ) anodes: Meso-scale modeling and deep learning with convolutional neural networks10.1016/j.egyai.2021.100122 ↗
35. Neural network-based modeling for a solid-oxide fuel cell stack
36. Data-driven prognostics of the SOFC system based on dynamic neural network models10.3390/en14185841 ↗
37. Control-oriented extraction and prediction of key performance features affecting performance variability of solid oxide fuel cell system
38. Fundamentals of recurrent neural network (RNN) and long short-term memory (LSTM) network10.1016/j.physd.2019.132306 ↗
39. On the use of neural networks and statistical tools for nonlinear modeling and on-field diagnosis of solid oxide fuel cell stacks10.1016/j.egypro.2014.01.032 ↗
40. Health state prediction and analysis of SOFC system based on the data-driven entire stage experiment10.1016/j.apenergy.2019.04.053 ↗
41. A comparative study of the Kalman filter and the LSTM network for the remaining useful life prediction of SOFC10.3390/en16093628 ↗
42. Attention is all you need
43. Bai · 2018
44. Miller · 2018
45. Detecting strange attractors in turbulence
46. Solid oxide fuel cell: Perspective of dynamic modeling and control10.1016/j.jprocont.2011.06.017 ↗
47. Neural network-based modeling for a solid-oxide fuel cell stack
48. Prognosis of fuel cell degradation under different applications using wavelet analysis and nonlinear autoregressive exogenous neural network10.1016/j.renene.2021.07.097 ↗
49. Simon · 2008
50. Early warning of tunnel collapse based on adam-optimised long short-term memory network and TBM operation parameters10.1016/j.engappai.2022.104842 ↗
51. Tiumentsev · 2019
52. 1D convolutional neural networks and applications: A survey10.1016/j.ymssp.2020.107398 ↗
53. Deep learning10.1038/nature14539 ↗
54. Tailoring the microstructure of a solid oxide fuel cell anode support by calcination and milling of YSZ10.1038/srep27359 ↗
55. Vafaeenezhad · 2022
56. Microstructure and long-term stability of Ni–YSZ anode supported fuel cells: a review10.1088/2752-5724/ac88e7 ↗
57. Modeling and microstructural study of anode-supported solid oxide fuel cells: Experimental and thermodynamic analyses
58. Chollet · 2021
59. Hierarchical delay-memory echo state network: A model designed for multi-step chaotic time series prediction10.1016/j.engappai.2021.104229 ↗
60. Hyperband: A novel bandit-based approach to hyperparameter optimization
61. Kingma · 2014
62. Jadon · 2022