CAJ | 학술논문

遗传算法中的交叉概率和变异概率是影响算法行为和性能的关键所在,直接影响算法的收敛速度,甚至影响有限进化代内的收敛性.本文通过分析交叉概率和变异概率对算法的影响,设计了一种依据种群多样性和进化代数自适应调节的交叉概率和变异概率,改善了传统遗传算法存在"早熟"现象和算法后期收敛速度慢的不足.最后,给出了三个典型函数的模拟例子,通过与传统SGA和AGA的对比结果显示,本文的改进提高了算法的性能.
유전산법중적교차개솔화변이개솔시영향산법행위화성능적관건소재,직접영향산법적수렴속도,심지영향유한진화대내적수렴성.본문통과분석교차개솔화변이개솔대산법적영향,설계료일충의거충군다양성화진화대수자괄응조절적교차개솔화변이개솔,개선료전통유전산법존재"조숙"현상화산법후기수렴속도만적불족.최후,급출료삼개전형함수적모의례자,통과여전통SGA화AGA적대비결과현시,본문적개진제고료산법적성능.
The keys which affect the genetic algorithms and convergence speed are mutation probability and crossover probability. To analyze their effect of algorithm, a new Self-adaptive mutation probability and crossover probability are designed according to diversi-ty in the population and generation number. The aim is to improve the defects of GA: easy prematurity and slow convergence speed in the rear of algorithm. Three typical function test s are given in this paper. By comparing with SGA and AGA, the result indicates the effectiveness of this improvement.