Parameter identification of fuel cell using Repairable Grey Wolf Optimization algorithm

Abstract

The performance of population-based meta-heuristic algorithms can be divided into two phases: exploitation and exploration. However, a major drawback of these methods is their inability to guarantee thorough searching of the entire search space. This is because the population (wolves) is randomly distributed and may not cover all areas in searching for the best solution. To address this issue, this paper introduces a simple technique called Repairable Grey Wolf Optimization (RGWO), which can be easily integrated into any population-based algorithm to enhance the exploration phase. The RGWO is implemented within the Grey Wolf Optimization (GWO) method and involves two main phases: elimination and search space management. In the elimination phase, underperforming wolves are gradually replaced with new ones, and new design parameters are introduced to improve the control of exploitation and exploration. In the search space management phase, instead of initially spreading the wolves across the entire search space, they are concentrated in a small portion and gradually expand from there. To evaluate the effectiveness of this approach, an extensive comparative study was conducted, comparing it with various classic and state-of-the-art meta-heuristic optimization algorithms. The results demonstrate several advantages over existing techniques, including fast convergence and high robustness.