Uso de algoritmo genético com operadores modificados para otimização de funções de variáveis reais

Abstract

and a novel statistic correlation mutation algorithm (CAM). Both ADX and CAM work with population information to improve existing individuals of the GA and increase the exploration potential via the correlation mutation. Solution-based methods offers good local improvement of already known solutions while lacking at exploring the whole search space, evolutionary algorithms provide better global search in exchange of exploitation power. Methods that increase the search potential are widely used for constrained optimization problems due to increased global and local search capabilities. The GA with the proposed operators improves results of constrained problems by balancing the exploitation and exploration potential of the algorithm. The conducted tests present average performance for various CEC’2015 benchmark problems, while offering good reliability and superior results on path planning problem for redundant manipulator and most of the constrained engineering design problems tested when compared with current works in the literature and classic optimization algorithms.