Coevolutionary algorithm for numerical optimization

To improve the global astringency of evolutionary algorithm,an algorithm based on multi-population concurrent evolution is proposed.According to the biological genetic diversity theory,the new algorithm maintains a single population’s relative purity and the whole population’s richness of breeding way.Different operators are used for different populations which breed in different habitats,so as to keep the diversity of population genes.When the algorithm is entrapped into a local optimal solution field,the reverse optimization can be applied to seek for its antithesis individuals to lift it out of the local optimal solution field.The experimental result shows that the new algorithm ranks first in all single and comprehensive items in comparison with a group of optimization data.