Fuzzy methods of driving genetic algorithms

This article presents two concepts of modified genetic algorithms, they employ a fuzzy logic controller to set a trend individuals´ evolution. In the algorithms we use a fuzzy logic controller, evaluating each individual as a parent for the next population. The fuzzy logic controller evaluates all individuals using fitness functions for earlier populations, which help´s to keep the knowledge collected in the prior populations. The controller modifies the probability of selection to parents´ pool, or probability of mutation, so in the fuzzy controlled genetic algorithms, a number of better quality individuals are larger then in the elementary genetic algorithms. We use the traveling salesman problem (TSP) as illustrations.