Abstract: In order to overcome premature convergence of traditional genetic algorithm (TGA), a multi-parent crossover GA (MPCGA) is proposed. Compared with greedy crossover, the proposed crossover can not only preserve the useful information of parent, but also maintain the diversity of population. In addition, the novel algorithm can also balance the trade-off between the exploitation and the exploration. Simulation results on traveling salesman problem (TSP) show the effectiveness and robustness of the novel multi-parent crossover GA.

Key words: multi-parent crossover, GA, diversity, traveling salesman problem (TSP)