Artur Gola

The methods goings-on with infeasible solutions in Genetic Algorithms' population

885

Abstract

GA are used to solve many optimalize problems, especially problems with constraints. The constraints bring that search space of the problem consists two disjoint sets: feasible space and infeasible space. Solving optimalize problem we look for a feasible optimal. There are feasible and infeasible individuals in search process. Finding a proper evaluation measure for feasible and infeasible individuals is very importance because it directly influences on the issue of the algorithm. This research approaches constraints problem in the GA. It contains also description of a new approach goings-on with infeasible individuals and its comparison with existing methods for example backpack loading problem.

Key words: genetic algorithm, measure of the fitness, infeasible solutions