具有生存期的开放式遗传算法

早熟是遗传算法应用中面临的主要问题,避免早熟现象的发生需要种群中的个体具有多样性。文章按进化过程中种群是否从外界引入新个体,将遗传算法分为开放式遗传算法和封闭式遗传算法,从外界引入个体可以增加种群的多样性,避免早熟。另外,当某个体参与交叉的次数达到一定阈值时(即生存期满),文中认为该个体已经充分地将其遗传基因传递到后代,该个体及其副本可以从种群中清除,从而避免出现某个体占领种群的现象,也能有效地保持种群的多样性。实验证明,将生存期与开放式种群结构相结合,可以显著提高遗传算法的收敛性能。

An Open Genetic Algorithm with Life-span

This paper presents an Open Genetic Algorithm(OGA)with lifespan.In nature,an animal stops producing de-scendents when it gets old and leaves the population when it dies.By simulating this propagating process,Genetic algo-rithms (GAs )try to find an optimum to an optimized problem.In Simple Genetic Algorithm(SGA),a chromosome with high fitness value may keep living for many generations,which is quite different from the natural process.The authors believe that lifespan is very important in making the world diversified and the population in GA can keep diversified if they confine the life span of chromosomes and give more chance to new individuals to compete.Further more,in SGA each chromosome can only crossover with another one in the same population,it maybe useful to introduce foreign chro-mosomes to take part in the process to improve the diversity of the population and prevent the population from con-verging to a local optimal solution.They also propose the Open Genetic Algorithm by combining the idea of lifespan and open population,and the following experiments have proved the advantage of OGA.