Study of the effect of the genetic parameters about the optimized solution and time of convergence in genetic algorithms with binary and real codification - DOI: 10.4025/actascitechnol.v25i2.2169

• Autor(a) principal: Catarina, Adair Santa
• Data de Publicação: 2008
• Outros Autores: Bach, Sirlei Lourdes
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Acta scientiarum. Technology (Online)
• Texto Completo: http://www.periodicos.uem.br/ojs/index.php/ActaSciTechnol/article/view/2169
• topic: algoritmos genéticos; parâmetros genéticos; codificação binária; codificação real; 3.00.00.00-9 Engenharias

This work presents a study about the effects of variation of genetic parameters over the solution quality and time of convergence in genetic algorithms that uses binary and real codification. Three algorithms was developed: the first called “AG Clássico” uses binary codification, the others two, called “AG M1” and “AG M2”, uses real codification. The parameters studied for “AG Clássico” was the population’s size, the number of individuals in selected elite and the mutation rate. The parameters studied for “AGM1” and “AG M2” was the population’s size, the number of individuals in selected elite, the number of individuals in random cloud generated around the individuals, the standard-deviation used to create the cloud and the mutation rate. The “AG Clássico” was affected by the parameters size of initial population, number of individuals in selected elite and mutation rate. The “AG M1” was affected only by the size of initial population. The “AG M2” was affected by the parameters size of initial population, number of individual in selected elite and mutation rate.