Abstract: An improved bee evolutionary genetic algorithm has been proposed.In the algorithm，based on the analysis of relationship between the random population scale and performance of convergence，it can be found that the random population scale is dynamically adjusted to the population structure.The well-phased control strategy is built up to adjust the random population scale.Because the scale of random population changes step by step，it can ensure the diversity of population，in the same time，it can also improve speed and accuracy of convergence.The simulation tests of the classical function show that the proposed algorithm is effective and feasible.

Key words: genetic algorithm, bee, random population scale, convergence

摘要： 提出了一种改进的蜜蜂进化型遗传算法。在该算法中，通过分析随机种群规模对算法收敛性能的影响，可以发现在算法的搜索过程中，对随机种群规模的需求是随群体状态的演变而动态变化的。为实现对随机种群规模的优化，提出使用分阶段调整的策略对随机种群规模进行动态调控，由于随机种群规模的渐进式变化，不但保证了种群的多样性，同时提高了算法的收敛速度和精度。对典型高维函数的优化实验结果表明了算法的有效性和可行性。

关键词: 遗传算法, 蜜蜂, 随机种群规模, 收敛性