一种遗传算法交叉算子的改进算法

An improved crossover operator of genetic algorithm

为了有效克服遗传算法收敛速度慢和易陷入局部极值点的缺点,提出了一种遗传算法交叉算子的改进算法,即采用自适应交叉概率,给不相关大的个体赋予较大的被选概率的配对方式进行交叉操作;在适应度比例轮盘赌的基础上辅以父子竞争的选择操作。二元多峰值Schaffer函数优化的仿真实例结果表明:与保留最优个体策略的遗传算法相比,改进算法能有效减少无效的交叉操作,收敛速度和全局搜索能力都得到了较大提高,其平均收敛代数和收敛到最优解的概率都优于保留最佳个体策略的遗传算法。

In order to effectively overcome the disadvantages of traditional Genetic Algorithm which converge slowly and easily run into local extremism, an improved crossover operator of genetic algorithms was proposed. This operator used the autoadaptive crossover probability and entrusted individual having big irrelevance index with a big elected probability to carry on the crossing operation; The two generations competitive selective operator was designed to improve the traditional genetic algorithm based on roulette. In a simulative example of multi-peaks function, the proposed method can reduce useless crossover effectively and thus the convergence speed and the search capability are greatly improved when compared with the elitist reserved genetic algorithm that keeps best strategy. As a result, the average convergence generations and the probability of getting optimal result are superior to the elitist reserved genetic algorithm.