在实际的多目标优化中，决策者通常只对少部分的Pareto最优解感兴趣。然而，传统多目标优化算法关注整个Pareto最优面上的解集，这不仅需要花费大量计算时间在无用解的搜索上，同时决策者也很难从众多解中选出符合自己偏好的解（特别是问题目标个数大于3时）。为此，本文提出了一种利用个体间的角度关系的偏好多目标进化算法。该方法通过重新定义个体间的支配关系和聚集距离使那些离决策者偏好区域越近的个体优先被保留下来，从而引导种群趋近于决策者的偏好区域。%The decision maker(DM)usually is only interested in small part of Pareto optimal solutions in the actual multi-ob-jective optimization .However ,the traditional multi-objective optimization algorithms pay more attention to the entire surface of the Pareto optimal solution set which need to spend large amount of computation time on the futile solutions search and the Decision Maker is difficult to select their preferred solutions from numerous solutions (especially when the problem’ s object is greater than 3 ) .Therefore ,this paper proposed a preference multi-objective evolutionary algorithm which is based on the angle information of in-dividuals .This method could priority retain these individuals which is closer to the DM’s preference region by redefining the domi-nation relation and crowding distance of individuals ,then the population will be guided to the DM’s preference region .
展开▼
