Solution of design optimization problems via metaheuristic search methods

Abstract

Metaheuristic algorithms inspired by natural phenomena are frequently used in solving optimization problems recently. Just as every problem has its characteristics, every algorithm has its unique structure. Therefore, problem-specific algorithm selection is an important issue. In addition, metaheuristic algorithms are very open to development. Therefore, improved/modified versions of algorithms are common. Working with benchmarking problems and engineering design problems is the best way to compare the performance and reliability of metaheuristic algorithms. In this study, the performances of firefly (FA), particle swarm optimization (PSO), bat algorithm (BA), ant colony optimization (ACO), glow worms (GSO), and hunting search (HuS) algorithms are compared. An enhanced version of the firefly algorithm (RFA) is also recommended and included in the comparison. A benchmark and five engineering design problems were selected for comparison purposes. All algorithms are set to twenty thousand iterations. The results show that the metaheuristic algorithm that gives the best results varies according to the nature of the problem. Moreover, although it does not change the ranking of the algorithm that gives the best result according to the problem, it shows that RFA gives better results in every problem than FA.