Genetic algorithm modification: addition of the population improvement stage

Abstract

The genetic modification of the algorithm is based on the introduction of a new stage consisting in improving the individual by inversion of the bit vector and eliminating the mutation stage. A necessary attribute of the improvement of the individual is the exceeding of the measure of the Hamming distance between the selected individuals by more than n/2, where n is the Hamming distance, which in the bit representation is the same as the number of genes in the individual. In this case, the "worse" the "non-ideal" individual is, the "better" it becomes after the inversion. Partially, this is compensated by the elimination of the mutation stage, and the overall effect in speed is achieved due to increasing the rate of convergence. The proposed improvements to the classical genetic algorithm allow increasing the convergence rate by 25-35% and the algorithm speed by 15-25%.