Using multiple selections to improve the efficiency of the genetic algorithm
Genetic algorithm (GA) is a popular optimisation method used in various fields. Although it has been used for quite some time, some of its disadvantages are still subject to research: premature convergence, computational expensiveness, and the setting of the initial parameters. This paper focuses on selections. Selections that can change parameters during the run of the GA have been developed, but a GA that uses different selections in individual generations has not been created. Since this approach yields good results for crossover and mutation, it raises the following question: could it also yield good results in the case of selection? The problem lies in the uniqueness of selections and the impossibility of using the same approach in combining them. Attempts to solve this problem have led to the modification in the GA structure which resulted in a new algorithm called “modified GA” (MGA). It turned out that the new structure can improve the efficiency of the GA.