自适应的多种群并行遗传算法研究

Research on Adaptive and Multipopulation Parallel Genetic Algorithm

典型的遗传算法本质上是一种并行的随机搜索策略,它不能对进化方向做出正确的感知和预测。文章定义了一个可以感知进化方向和衡量进化速度的指标,指导遗传算子做出自适应的调整。结合并行计算和多种群进化思想,提出了“带环多种群模型”,并构造了一种自适应的多种群并行遗传算法。实验结果表明,该算法可以有效地引导和控制进化方向,克服过早收敛现象,提高搜索效率。

In substance,Genetic Algorithm is a parallel and stochastic search game,so it can not availably apperceive and forecast the orientation of evolution.To solve the problem,an index that can reflect the orientation and pace of evolution in this paper is defined.The index is used to guide the evolution and to adaptively adjust genetic operator.Based on the idea of parallel computing and multipopulation evolution game,the paper also presents a simplex annulus and multipopulation parallel model and constructs an adaptive and multipopulation parallel genetic algorithm.The experimental results indicate the algorithm can be used availably to guide and adjust evolution,and to improve the search efficiency.