Metaheurística Inspirada no Comportamento do Corvo com Multiestratégias de Aprendizagem Baseada em Oposição

Abstract

Frequently, metaheuristics algorithms are used in many scientificand engineering problems optimization. However, these metaheuristicscome across an impossibility to escape from optimal solutions.To equilibrate the balancing between exploitation and exploration,several proposals, inspired in the behavior of animalgroups, have been developed. A recent metaheuristic is Crow SearchAlgorithm (CSA). It is inspired in the intelligent behavior ofcrows. This paper presents a novel CSA variant combined withtwo strategies opposed to based-learning called Elite Opposition-Based Learning and Generalized Opposition-Based Learning. Toevaluate the performance of the proposed algorithm we executedtwo experiment groups that utilized twelve reference functions. Inthe first group, the proposal was compared to a CSA variant andother three algorithms. For the second experiment group, we comparedto other four algorithms and more one variant. After the executionof experiments, we observed that the obtained numerical resultsshowed a superiority when compared to the other algorithms.