基于纯数值函数优化的一种混合遗传算法

A Kind of Hybrid Genetic Algorithm Based on Pure Numerical Function Optimization

摘要: 通过引入与进化代数相关的交叉概率和与个体适应度相关的变异概率的自适应遗传算子同时把Powell局部寻优算法融入遗传算法的搜索过程构成了一种数值函数全局寻优的混合遗传算法.实验表明混合遗传算法改善了遗传算法的局部搜索能力,有效地解决了遗传算法的早熟现象,显著提高了遗传算法求得全局解的概率.同时由于混合遗传算法中只利用函数值信息,所以该混合遗传算法是纯数值函数的优化的一种通用方法.

Abstract: A hybrid genetic algorithm for locating the global optimization of numerical function is put forward by designing self-adaptive genetic operators and setting the Powell algorithm in genetic algorithm.The hybrid approach improves the local searching ability to the genetic algorithm and promotes the probability for the global optimization greatly.Because only the objective values are used,the hybrid approach is a generalized algorithm for numerical functions.