Current Proceedings on Technology
Yazarlar: Hemmak Allaoua, Bouderah Brahim
Konular:-
Anahtar Kelimeler:Genetic algorithm,Crossover,Mono crossover,Combinatorial optimization,TSP.
Özet: In this paper, we will present a survey of some mono crossovers methods which can be used to product successive generations in genetic algorithm (GA) process. We will apply them to the Travelling Salesman Problem (TSP) for using its available library TSPLIB. The introduced idea, inspired from reproduction phenomenal of some spices, consists to use only one parent to make just one child in crossover phase of GA. So, we will propose four versions and compare them to a pair parents case to show, on one hand, that mono crossover can give better solution than ordinary pair crossover which can often lead to lethal genes, on the other hand, the version that simulate nearly the natural process where the child may almost acquire his grandfather characters, is the best one. Also we will empirically look for the ideal ratio of parent’s gene that may participate to crossover operator.