Статья

AI-assisted multi-objective optimization of Ni anode microstructure for enhanced electrochemical performance and long-term stability of solid oxide fuel cells

Mohamed ShabanPhysics Department, Faculty of Science, Islamic University of Madinah, P. O. Box: 170, Madinah 42351, Saudi ArabiaAs’ad AlizadehDepartment of Civil Engineering, College of Engineering, Cihan University-Erbil, Erbil, IraqRashed Abu HammourFaculty of Technical Education, Hourani Center for Applied Scientific Research (HCASR), Al-Ahliyya Amman University, Amman, JordanKamal SharmaDepartment of Mechanical Engineering, GLA University, Mathura, IndiaSafi AlshammariDepartment of Mechanical Engineering, College of Engineering, Jouf University, Sakakah 72388, Saudi ArabiaMehraj‐ud‐din NaikDepartment of Chemical Engineering, College of Engineering and Computer Science, Jazan University, Jazan, Saudi ArabiaAbdellatif M. SadeqFaculty of Agricultural Mechanization, TIIAME National Research University, Kori Niyoziy 39, Tashkent 100000, UzbekistanNarinderjit Singh Sawaran SinghCentre of Research Impact and Outcome, Chitkara University, Rajpura 140417, Punjab, IndiaHusam RajabCollege of Engineering, Department of Mechanical Engineering, Najran University, King Abdulaziz Road, P.O Box 1988, Najran, Saudi ArabiaRoshan SinghDivision of Research and Development, Lovely Professional University, Phagwara, Punjab, India

2026en

ABI

Аннотация

Nickel (Ni) agglomeration in porous anodes is a major microstructural degradation mechanism that progressively deteriorates the electrochemical performance and long-term durability of anode-supported solid oxide fuel cells (SOFCs). During prolonged operation, Ni particle coarsening alters the effective three-phase boundary, electronic conductivity, mass transport, and percolation characteristics of the anode, thereby creating a complex trade-off between initial power generation and degradation resistance. This study develops an artificial intelligence-assisted multi-objective optimization framework to identify Ni anode microstructures that simultaneously enhance power density and mitigate agglomeration-induced degradation. A validated numerical dataset describing the long-term behavior of an anode-supported SOFC is employed, with the initial Ni particle diameter (D Ni,i ), particle size ratio (PSR), and Ni solid-phase volume fraction (VF) considered as the principal microstructural variables. A three-stage framework integrating neural-network surrogate modeling, multi-objective optimization, and multi-criteria decision-making is developed. First, genetic algorithm-optimized multilayer perceptron neural networks (GA-MLPNN) and elk herd optimizer-based MLPNN (EHO-MLPNN) models are evaluated for predicting average power density (APD) and degradation rate (DR). The GA-MLPNN model demonstrates superior predictive performance, achieving a correlation coefficient of R = 0.9984 compared with 0.9912 for EHO-MLPNN. Subsequently, multi-objective chaos game optimization (MOCGO) generates a Pareto-optimal set that reveals the inherent trade-off between electrochemical power output and long-term degradation. The optimized solutions provide APD values of 2602–2772 W/m 2 and DR values ranging from −0.33 to 0.16%/1000 h. The results indicate that smaller initial Ni particles (~0.60 μm), relatively high PSR values, and high Ni volume fractions (approximately 0.53–0.55) are generally associated with favorable performance–durability combinations. Finally, the MARCOS method identifies preferred solutions under seven different weighting scenarios, providing flexible design strategies according to the relative importance assigned to power generation and durability. The proposed framework demonstrates the potential of artificial intelligence and multi-objective decision analysis for optimizing Ni-containing SOFC anode microstructures and mitigating the performance degradation associated with long-term Ni agglomeration.

Перевод пока недоступен

Идентификаторы

DOI: 10.1016/j.elecom.2026.108256

Цитирования и источники

Цитирований: 0Использованных источников: 48