Examination of parameters used in ant colony algorithm over truss optimization

Abstract

Metaheuristic optimization techniques have been used to solve engineering problems with an increasing speed for the last 30 years. Most of these algorithms have been developed by imitating a process in nature. In this study, the ant colony algorithm inspired by the natural life of ants is discussed. The ant colony algorithm requires some parameters to perform an optimization, as in other meta-heuristic algorithms. The aim of this study is to examine the effect the values of the parameters used in the ant colony algorithm on the results. For this purpose, as an exemplary problem, a study was carried out on the optimization of truss systems, one of the constrained problems frequently discussed in the literature. Appropriate values of optimum design parameters such as number of ants, pheromone update coefficient and penalty coefficient were investigated using the coded computer program. As a result of the study, the effect of the relevant parameters on the result was determined and the points to be considered in the selection of these parameters were specified.