Artificial neural network modeling and optimization of the Solid Oxide Fuel Cell parameters using grey wolf optimizer. (November 2021)

• Record Type: Journal Article
• Authors: Chen, Xinxiao; Yi, Zhuo; Zhou, Yiyu; Guo, Peixi; Farkoush, Saeid Gholami; Niroumandi, Hossein
• Is Part Of: Energy reports. Volume 7(2021)
• Page Start: 3449
• Page End: 3459
• Publication Date: 2021-11
• Subjects: Solid oxide fuel cell -- Parameter identification -- Grey Wolf Optimizer -- Artificial neural network -- Performance improvement; Power resources -- Periodicals; Energy industries -- Periodicals; Power resources; Periodicals; Electronic journals; 621.04205
• DOI: 10.1016/j.egyr.2021.05.068
• Languages: English
• ISSNs: 2352-4847

Abstract

Abstract: Using Green and carbon-free energy sources is a new concept in the energy conversion, power generation, and energy management framework. Since there is a relatively small number of neural network applications in the field of fuel cells, especially in the case of solid oxide fuel cells, this work adopts the Artificial Neural Network model for modeling aims according to the empirical datasets. Besides, a new optimization method is applied to optimize the solid oxide fuel cell efficiency. The grey wolf optimizer with fast, robust, and simple features is applied to obtain the optimal operational variables of solid oxide fuel cells. The key operational parameters used for the optimization comprise the thickness of the anode support layer, the porosity of the anode layer, the thickness of the electrolyte layer, and the thickness of the cathode layer. The modeling results compared to the laboratory that confirms the ability of the artificial neural network model and optimization method in parameter identification. Two case study optimization procedure was assessed. Firstly, the variables optimized under the operational temperature of 800 °C and the values of 19 μ m, 0.52 mm, 62.16 μ m, and 75% are obtained for the electrolyte layer thickness, anode support layer thickness, cathode thickness, and anode support layer porosity, respectively. For the second case study, the power density based on the suggested method maximized up to 28% compared to the experimental results.