Baldinelli,A.Barelli,L.Bidini,G.Bonucci,F.Iskenderoğlu,F.C.2024-10-062024-10-0620170978-888286324-1[SCOPUS-DOI-BELIRLENECEK-106]2-s2.0-85173569758https://hdl.handle.net/20.500.14411/9469Consiglio Nazionale delle Ricerche; ITAE; The Bioelectrochemical Society (BES)For their high flexibility of operation, Solid oxide fuel cells (SOFCs) are promising candidates to coach the transition towards cleaner and efficient energy generation. Yet, SOFC performance might be markedly affected by fuel composition variability and operative parameters. For that, a reliable simulation tool is necessary for SOFC performance, to optimize its working point and to provide a suitable control. Given the high variability ascribed to the fuel and the electrochemical system high nonlinearity, the implementation of artificial intelligence techniques, like Artificial Neural Networks (ANNs), is worth considering. In ANNs, the correlation between system inputs and outputs is handled by virtual neurons, establishing in-out correlations without entering in knotty kinetics and material properties issues. For what above, a suitably sized experimental campaign is to be designed to provide a large data-set. This to guarantee high ANN performance in the voltage estimation and, at the same time, a wide application domain of the neural simulator. © EFC 2017 - Proceedings of the 7th European Fuel Cell Piero Lunghi Conference.eninfo:eu-repo/semantics/closedAccessArtificial Neural NetworksControl algorithmsLow-carbon fuelsSimulationSolid Oxide Fuel CellsConference ObjectN/AN/A99100