A comprehensive survey of artificial intelligence-based techniques for performance enhancement of solid oxide fuel cells: Test cases with debates

Abstract

Since installing solid oxide fuel cells (SOFCs)-based systems suffers from high expenses, accurate and reliable modeling is heavily demanded to detect any design issue prior to the system establishment. However, such mathematical models comprise certain unknowns that should be properly estimated to effectively describe the actual operation of SOFCs. Accordingly, due to their recent promising achievements, a tremendous number of metaheuristic optimizers (MHOs) have been utilized to handle this task. Hence, this effort targets providing a novel thorough review of the most recent MHOs applied to define the ungiven parameters of SOFCs stacks. Specifically, among over 300 attempts, only 175 articles are reported, where thirty up-to-date MHOs from the last five years are comprehensively illustrated. Particularly, the discussed MHOs are classified according to their behavior into; evolutionary-based, physics-based, swarm-based, and nature-based algorithms. Each is touched with a brief of their inspiration, features, merits, and demerits, along with their results in SOFC parameters determination. Furthermore, an overall platform is constructed where the reader can easily investigate each algorithm individually in terms of its governing factors, besides, the simulation circumstances related to the studied SOFC test cases. Over and above, numerical simulations are also introduced for commercial SOFCs’ stacks to evaluate the proposed MHOs-based methodology. Moreover, the mathematical formulation of various assessment criteria is systematically presented. After all, some perspectives and observations are provided in the conclusion to pave the way for further analyses and innovations.