论文部分内容阅读
本文在经典遗传算法中增加了一个称之为空间交配的遗传算子,通过空间交配,把整个搜索空间划分成不相交(或近似不相交)的子空间,它不但优化空间内的个体,而且各子空间作为一种运算对象,其本身也得到不断优化.算法不是在某一时刻才把群体进行突发多样性化,而是群体在总体上时刻保持多样性,同时不断积累有用信息,最终使算法收敛到全局最优解,而且子群体的个体交换很少,解决了现在大多并行遗传算法的系统通信开销大的问题。
In this paper, we add a genetic operator called space mate to classical genetic algorithm, and divide the entire search space into disjoint (or nearly disjoint) subspaces through space mating. It not only optimizes individuals in space, but also Each subspace as a computing object, itself has been continuously optimized.Algorithm does not only burst the population at a certain moment diversity, but the population in general to maintain diversity at the same time continue to accumulate useful information, and ultimately Make the algorithm converge to the global optimal solution, and the individual exchange of subgroups is very little, which solves the problem of large system communication overhead of most parallel genetic algorithms.