基于隔离小生境的自适应遗传算法研究

A study of genetic algorithms based on isolated niche technique

基本遗传算法存在着收敛速度慢、易陷入局部最优等缺陷.针对这些缺陷,本文通过对自然界中隔离机制和自适应技术的研究,发现隔离机制对于改善种群的多样性等方面有良好的作用,而交叉概率和变异概率不仅与自身的适应度值有关而且同进化代数有较大关系.本文结合隔离小生境的知识和自适应技术提出了一种基于隔离小生境技术的自适应遗传算法,并通过函数测试表明该方法能够有效改善种群多样性,提高算法的全局寻优能力和收敛速度.

Simple genetic algorithms get local optimal solution too easily and converge slowly. To solve these problems, isolation mechanism and self adapting technique are studied in this article, and we find isolation mechanism plays important roles in evolution and can actively improve the diversity of populations, while crossover rate and mutation rate are not only related to fitness function value but also affected greatly by evolution eras. In this article, a new adaptive genetic algorithms based on isolated niche technique is presented, and classical and practical examples simulated. The results show that the adaptive genetic algorithms based on niche technique has greatly improve the diversity of populations, quicken the speed of convergence and enhance the probability of getting the best solution.