Overcoming a recent impasse in the application of artificial neural networks as solid oxide fuel cells simulator with computational topology

Abstract

In recent years, the solid oxide fuel cell (SOFC) scientific community has invested continuous efforts to employ artificial intelligence methods to design and develop new energy systems. It is crucial to gain a better understanding of the microscale phenomena that occur in the electrodes. In this review, we present a literature review of the field, discussing the limitations of including microstructural data in existing research and possible research directions to overcome them. This review focuses on a particular research area that uses artificial neural networks (ANNs) to predict the performance of SOFCs. Herein, we show that neural networks are used not only to conform to the newest trends but also for improving the design and providing a better understanding of microscale phenomena that occur in the electrodes. The review concludes by highlighting topological data analysis as a promising area of research that can incorporate detailed microstructure characterization in ANNs for performance prediction.