Estudio de estrategias de reemplazo de roles en la Optimización basada en Lobos Grises

Abstract

Improving the behavior of metaheuristic algorithms has been, is and will be a challenge for the scientific community. Strategies aimed at improving exploration of the search space and avoiding stagnation of solutions are some of the most studied premises in the literature. The Gray wolf-based optimization (GWO) metaheuristic is capable of solving continuous optimization problems by applying a command role assignment scheme that provides an adequate balance between exploration and intensification of solutions. In this article, we will analyze some strategies for defining roles in GWO and measure their influence on the quality of the search process in a continuous space. For strategies, a probabilistic selection method is used, by distance followers and a combination of both. The experimental results showed that the follower-based variants provide greater stability in the results, and in addition, the probability-based models present greater effectiveness under a probability value of 0.9.