Tribe-charged system search for global optimization

Abstract

In this paper, an improved metaheuristic algorithm called Tribe-Charged System Search is proposed for global optimization. In this improved algorithm, the main searching loop of the standard Charged System Search algorithm is modified to achieve a better convergence performance. In this algorithm, the searching phase of the algorithm is divided into three distinct phases called "Tribes" in which the searching process is conducted differently in each phase based on the communication allowance between tribes. These changes cause the algorithm focuses on high global searching for the early iterations while the concentrating in local search process is considered through the last iterations. In order to validate the performance of the new algorithm, 4 mathematical benchmark functions alongside the 3 well-known constrained problems and 2 engineering design problems are utilized. The performance of the new algorithm is compared to six other metaheuristic algorithms. The results of the mathematical, constrained and engineering problems prove that the new method is capable of providing very competitive results among the other metaheuristics.