通过对浮点遗传算法早熟收敛现象的分析 ,提出了一种新的父代选择策略 ,即使用当前代的子代个体作为下代的父代个体 ,可使交叉算子持续地探索和开发新空间。引入对个体的代数保护策略 ,即在它发生变异前保证有足够的演化 ,可以避免对新空间不成熟的开发。通过与其它父代选择策略的对比 ,并通过实验和 GENOCOP系统比较 ,表明本方法能得到较好的结果。 By analyzing the phenomenon of premature convergence in floating-point genetic algorithm, a new parent selection strategy is proposed. That is, using the current generation of offspring as the next generation of parents can make crossover operators continuously explore and develop new space. The introduction of an algebraic protection strategy for individuals ensures that there is sufficient evolution before it occurs to avoid the immature development of the new space. By comparing with other parental selection strategies and comparing with the GENOCOP system through experiments, it shows that this method can get better results.