Abstract: The simple genetic algorithms have characteristic of premature convergence，long time of optimization and poor performance of local searching ability.Their computational defects are analyzed.Based on elitist strategy，immigration strategy and fitness average of population，the novel adaptive genetic algorithms is proposed.The convergence of the novel self-adaptive genetic algorithms is proved.The results of a simulation example show that the proposed self-adaptive genetic algorithms have an obvious advantage of higher searching efficiency and accuracy and overcoming premature convergence.

Key words: genetic algorithm, premature, immigration strategy, elitist strategy, fitness average

摘要： 针对基本遗传算法具有早熟性收敛、寻优时间长及局部搜索能力差的问题，分析产生这些问题的原因。结合最优保存策略和移民策略，提出基于种群平均适应度信息的遗传算法自适应算子的改进方案，并对改进遗传算法的收敛性予以证明。仿真结果表明，改进遗传算法在搜索效率、搜索精度和克服早熟收敛现象方面均有明显的优越性。

关键词: 遗传算法, 早熟性收敛, 移民策略, 最优保存策略, 平均适应度