Abstract: After analyzing the premature convergence problem of genetic algorithm based on standard selection，crossover and mutation operations，this paper proposes a modified genetic algorithm with natural cognation exclusion，which exploits the individual’s married recording and inheriting to avoid inbreeding，and changes the crossover and mutation probability self-adaptively in according to the population information entropy.New algorithm resolves the premature convergence problem perfectly by keeping population diversity and important building block.The experimental results also show that new algorithm has the performance of global convergence.

Key words: natural cognation, population diversity, genetic algorithm, premature convergence, population information entropy

摘要： 分析了选择、交叉和变异操作下遗传算法早熟收敛问题，提出了一种具有自然血亲排斥的遗传算法，利用个体通婚记录信息和婚史继承方法有效地避免了三代自然血亲之间的近亲繁殖，并依据群体信息熵实现交叉率和变异率的自适应变化以及交叉变异位置的自适应确定，保证了群体的多样性和重要构造块的保护，扩大搜索空间，较好地解决了早熟收敛问题，从而实现全局收敛。仿真结果表明，所提出算法有较好的全局收敛能力。

关键词: 自然血亲, 群体多样性, 遗传算法, 早熟收敛, 群体信息熵