A study on novel genetic algorithm with sustaining diversity

Traditional crossover and mutation have difficulties in searching the optimum individual because of the random selection of gene positions. In the paper, the impact of the crossover position and mutation position upon the diversity in the population is analyzed. Then the concept of efficient crossover position length is introduced and some important conclusions are given. Finally, a new kind of diversity-maintaining mutation is proposed, of which the mutation probability and position are varied depending on the gene diversity. In the rest of the paper a series of experiments are made to verify our ideas and sound results are achieved.