Gender genetic algorithm with learning in the dynamic optimization problem

Abstract

The article analyses an approach to the optimization of fast-changing processes using a gender-based genetic algorithm. The suggested algorithm differs from the traditional genetic algorithm in that it divides an artificial population into two groups according to their gender. This separation allows combining the rapid adaptability of the population resulting from variations in the male subpopulation with the fixed adaptability of the female population. The article demonstrates that the meta-learning effect based on the mutation parameters in combination with individual learning, i.e. the Baldwin effect, is more effective for finding dynamic optimal solutions than the classic gender-based genetic algorithm and the Lamarckian genetic algorithm. The dynamics of extinguishing natural fires is considered as a promising application of the developed gender-based genetic algorithm with the Baldwin effect.