遗传算法的有效基因块保护策略*

遗传算法是一种结合全局搜索和局部搜索两种特性的自适应搜集随机算法,但存在早熟性收敛和收敛速度慢两方面问题。由于遗传算法运行过程中最小诱导模式普遍存在于个体中,同时在遗传算法运行后期,个体中存在很多属于收敛优化解或全局最优解的基因块。通过分析和论证,建立了保护属于最小诱导模式或优化解的有效基因块的控制策略。该策略可与其他杂交算子和变异算子结合,为遗传操作中父代个体包含的非有效基因块基因座上的基因提供更多进化机会,从而提高这些基因座上的有效基因数量,维持有效的种群多样性,较好地抑制了GA的早熟现象,提高了算法收敛速度和全局寻优能力。

Protect strategy for effectual gene block of genetic algorithm

Genetic algorithms is a kind of self-adaptive stochastic search algorithms that combine global search and local search property,but there are two problems: premature convergence and slow-footed convergence.It was because the least induced schema was ubiquitous in individuals and individuals included many gene blocks belonging to convergence optimal solution or global optimal solution.The protect strategy for effectual gene block was established through analyses and argumentation.The strategy combining with other crossover operators and mutation operators provided more evolution opportunity for those genes locating on the locus included by non-effectual gene block of parents' individuals during genetic operation.Accordingly,the quantity of effectual gene on these loci was increased,and the diversity of population was hold in high quality,and the premature convergence was restrained preferably,and the convergence rate and the global seek-optima capability were improved.