基于Species机制的多目标遗传算法

多目标优化算法设计正在成为当前进化算法领域的一个研究热点.考虑将最初用于多峰优化的Species机制引入到多目标遗传算法中，通过借鉴现有多目标算法的相关思想，设计并提出了一种新的Species方法，包括基于Pareto最优性和拥挤度思想的Species种子确定策略及适应性的Species构造策略.一组标准多目标测试函数的仿真实验结果表明，提出的基于Species机制的多目标遗传算法表现出比经典的非支配排序遗传算法Ⅱ(NSGA Ⅱ)更好的性能.

Species based Genetic Algorithm for Multiobjective Optimization Problems

Considering that evolutionary multiobjective optimization has been becoming one of research topics in evolutionary algorithm community recently， a species based mechanism was introduced into multi objective GA， which is initially applied into GA for multi modal optimization problem. And then a new species based method was designed and proposed for multiobjective optimization problem， which comprises a species seed indentifying strategy based on Pareto optimality and crowd degree and an adaptive species constructing scheme. Experimental results showed that the proposed species based multiobjective GA outperforms nondominated sorting GA Ⅱ(NSGA Ⅱ)on a set of benchmark test problems.