一种求解多目标优化问题的新遗传算法

给出了种群个体的序和密度定义,在此定义下又给出目标空间中Pareto界面上解的质量度量函数和均匀性度量函数.把对任意多个目标优化问题转化成了利用解的质量度量函数及均匀性度量函数构成的2个目标优化问题,同时对转化后的优化问题设计了一种新的遗传算法.为了提高算法的性能,对算法采用变适应度进化策略及遗传参数采用自适应调节技术.数据实验表明该算法对提高Pareto界面上解的质量及其均匀分布是有效的.

A new genetic algorithm for multi-objective optimization

The rank and density of the population were firstly defined, and a quality measuring function of solutions and a uniformity measuring function of solutions in objective space were clearly given. The multi-objective optimization problem was finally converted into a two objectives optimization problem which was based on the quality measuring function of solutions and the uniformity measuring function of solutions. For the transformed problem, a novel genetic algorithm was proposed. In order to enhance the quality of the algorithm, the changing adjustment strategy of evolutionary was applied and the self-adjustment method was adapted into the parameter. The specific experiment shows that the algorithm is effective in dealing with the quality and the uniform distribution of Pareto solutions.