The fuzzy genetic strategy for multiobjective optimization

This paper presents the idea of fuzzy controlling of evolution in the genetic algorithm (GA) for multiobjective optimization. The genetic algorithm uses the Fuzzy Logic Controller (FLC), which manages the process of selection of individuals to the parents' pool and mutation of their genes. The FLC modifies the probability of selection and mutation of individuals' genes, so algorithms possess improved convergence and maintenance of suitable genetic variety of individuals. We accepted the well-known LOTZ problem as a benchmark for experiments. In the experiments we investigated the operating time and the number of fitness function calls needed to finish optimization. We compared results of the elementary algorithms and the modified algorithm with the modification of probability of selection and mutation of individuals. Some good results have been obtained during the experiments.