Artificial neural network simulator for SOFC performance prediction
(2002) In Journal of Power Sources 112(1). p.54-60- Abstract
- This paper describes the development of a novel modelling tool for evaluation of solid oxide fuel cell (SOFC) performance. An artificial neural network (ANN) is trained with a reduced amount of data generated by a validated cell model, and it is then capable of learning the generic functional relationship between inputs and outputs of the system. Once the network is trained, the ANN-driven simulator can predict different operational parameters of the SOFC (i.e. gas flows, operational voltages, current density, etc.) avoiding the detailed description of the fuel cell processes. The highly parallel connectivity within the ANN further reduces the computational time. In a real case, the necessary data for training the ANN simulator would be... (More)
- This paper describes the development of a novel modelling tool for evaluation of solid oxide fuel cell (SOFC) performance. An artificial neural network (ANN) is trained with a reduced amount of data generated by a validated cell model, and it is then capable of learning the generic functional relationship between inputs and outputs of the system. Once the network is trained, the ANN-driven simulator can predict different operational parameters of the SOFC (i.e. gas flows, operational voltages, current density, etc.) avoiding the detailed description of the fuel cell processes. The highly parallel connectivity within the ANN further reduces the computational time. In a real case, the necessary data for training the ANN simulator would be extracted from experiments. This simulator could be suitable for different applications in the fuel cell field, such as, the construction of performance maps and operating point optimisation and analysis. All this is performed with minimum time demand and good accuracy. This intelligent model together with the operational conditions may provide useful insight into SOFC operating characteristics and improved means of selecting operating conditions, reducing costs and the need for extensive experiments. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/324417
- author
- Arriagada, Jaime LU ; Olausson, Pernilla LU and Selimovic, Azra LU
- organization
- publishing date
- 2002
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- neural network, (ANN), artificial neural network, fuel cell, solid oxide fuel cells (SOFC), backpropagation, feed-forward network
- in
- Journal of Power Sources
- volume
- 112
- issue
- 1
- pages
- 54 - 60
- publisher
- Elsevier
- external identifiers
-
- wos:000179018400008
- scopus:0037167997
- ISSN
- 1873-2755
- DOI
- 10.1016/S0378-7753(02)00314-2
- language
- English
- LU publication?
- yes
- id
- 465e42c7-ab22-49e7-bc7d-562039bfeac0 (old id 324417)
- date added to LUP
- 2016-04-01 12:16:42
- date last changed
- 2022-03-28 22:40:40
@article{465e42c7-ab22-49e7-bc7d-562039bfeac0, abstract = {{This paper describes the development of a novel modelling tool for evaluation of solid oxide fuel cell (SOFC) performance. An artificial neural network (ANN) is trained with a reduced amount of data generated by a validated cell model, and it is then capable of learning the generic functional relationship between inputs and outputs of the system. Once the network is trained, the ANN-driven simulator can predict different operational parameters of the SOFC (i.e. gas flows, operational voltages, current density, etc.) avoiding the detailed description of the fuel cell processes. The highly parallel connectivity within the ANN further reduces the computational time. In a real case, the necessary data for training the ANN simulator would be extracted from experiments. This simulator could be suitable for different applications in the fuel cell field, such as, the construction of performance maps and operating point optimisation and analysis. All this is performed with minimum time demand and good accuracy. This intelligent model together with the operational conditions may provide useful insight into SOFC operating characteristics and improved means of selecting operating conditions, reducing costs and the need for extensive experiments.}}, author = {{Arriagada, Jaime and Olausson, Pernilla and Selimovic, Azra}}, issn = {{1873-2755}}, keywords = {{neural network; (ANN); artificial neural network; fuel cell; solid oxide fuel cells (SOFC); backpropagation; feed-forward network}}, language = {{eng}}, number = {{1}}, pages = {{54--60}}, publisher = {{Elsevier}}, series = {{Journal of Power Sources}}, title = {{Artificial neural network simulator for SOFC performance prediction}}, url = {{http://dx.doi.org/10.1016/S0378-7753(02)00314-2}}, doi = {{10.1016/S0378-7753(02)00314-2}}, volume = {{112}}, year = {{2002}}, }