一种新型保持种群多样性的遗传算法

A Modified Genetic Algorithm with Maintaining Diversity

针对遗传算法的“早期收敛”或“遗传漂移”,本文提出了一种新型保持群体多样性的遗传算法。该算法利用种群的熵和个体基因座的多样度来测度进化中种群的多样性,并将遗传操作与种群多样性联系起来,建立了遗传操作与种群多样度之间的函数关系式,使遗传算子中的主要参数能够随多样性函数和适应度函数的变化而自适应调整,从而提高了进化中种群的多样性和算法的搜索效率。仿真研究证明了该算法的优越性。

This paper presents a modified genetic algorithm with maintaining diversity for solving genetic excursion and premature convergence. In the algorithm, entropy of population and locus diversity take as the measure of population diversity of the evolving state. The relation between genetic operation and population diversity is set by their function, which makes the main parameter of genetic operation adaptively adjust according to the varying fitness function and diversity function. The algorithm greatly improves population diversity and the searching efficiency. The experiments show advantages of the algorithm.