Özet:

Metaheuristic based artificial intelligence algorithms are commonly used in the solution of optimization problems. Another area -besides engineering systems- where chaos theory is widely employed is optimization problems. Being applied easily and not trapping in local optima, chaos-based search algorithms have attracted great attention. For example, it has been reported that when random number sequences generated from different chaotic systems are replaced with parameter values in bioinspired and swarm intelligence algorithms, an increase in the performance of metaheuristic algorithms is observed. Many scientific studies on developing hybrid algorithms in which metaheuristic algorithms and chaos theory are used together are already in process. In this article, scientific studies that cover the most popular metaheuristic algorithms in the literature in recent years and chaos theory subtopics together are examined. Great number of studies on metaheuristic algorithms and chaos issues exist in scientific literature. This article, hitherto, had to be limited to the most common meta-heuristic algorithms. In chaos-based metaheuristic algorithms, some advantages such as easy implementation, short application time and search acceleration have been addressed. This article is believed to contribute to two groups of researchers: The first group includes researchers already using meta-heuristic algorithms and who will come to understand that they can even improve their current techniques with chaos theory subarguments. The second group includes those who currently utilize chaotic analysis and methods in areas like nonlinear prediction modeling design and who will realize that they can make their existing methods even smarter with metaheuristic methods. Some metaheuristic algorithms use chaotic maps to solve problems such as trapping in local optimal solutions and premature convergence. In this study, such algorithms are examined through the benchmark functions they use. In addition, PSO, FA, ABC, WO algorithms are compared in terms of their common features, and algorithms with the best success rate are presented.

Anahtar Kelimeler: