A new hybrid genetic algorithm based on clan competition

Abstract

It is difficult for basic genetic algorithm and evolutionary programming algorithm to converge at global optimal solution of real-continual function in practice, although both of the algorithms have the ability for getting the global optimal solution with convergent probabilities 1 in theory. In this paper, a new hybrid genetic algorithm based on clan competition is proposed, and it is proved that the probability of the new algorithm convegent to the global optimal solution is 1, Numerical experiments results illustrate that, compared with the former two algorithms, the new algorithm is the robustest among the three algorithms, what's more, it has the highest precision with the equal parameters.