中文摘要：

讨论了遗传算法在问题求解中的早熟现象,引进一个参数用以衡量种群中染色体的相似程度,用以增加种群的多样性;在杂交和变异运算过程中,混合了模拟退火思想作为新个体的接受准则;通常的变异算子需要扫描每一个染色体中每一个等位基因,提出一种新的变异方式,大大提高了算法搜索效率。通过实际计算比较表明,该改进遗传算法在背包问题求解中具有很好的收敛性、稳定性和计算效率。

英文摘要：

The paper discusses the premature convergence of canonical genetic algorithm.A parameter is introduced to weigh the chromosome similarity and increase the population diversity of the algorithm.The idea of simulated annealing is used to accept the new individual in the crossover and mutation operation.A new mutation operation is provided to improve the search efficiency.Experiments on knapsack problem illustrate that the new proposed genetic algorithm has better convergence,stability and efficiency.