一种基于邻居支配关系的偏好多目标进化算法

摘要

传统多目标优化算法得到的解集是整个Pareto最优面，需要花费大量精力在Pareto最优解的搜索上，同时当问题目标个数较多时，决策者很难从大量的解中选出自己最满意的解。因此，针对上述问题，提出一种基于邻居关系的偏好多目标进化算法。该算法通过一个邻居支配关系对非支配个体集进行适应度分层，借助参考点引导个体种群向决策者感兴趣的区域靠近。通过与几种经典的偏好多目标进化算法进行比较实验，结果表明，所提出的算法能引导种群趋近于决策者最满意的区域。%The solutions set gained by traditional multi-objective optimisation algorithms is the entire optimal surface of Pareto , this has to pay much attention to searching the Pareto-optimal solutions .Meanwhile , when the number of objectives for the problem is large , it’ s difficult for the decision maker to choose the most satisfying solution from so many solutions .Therefore, considering the above problems , we propose a neighbour relationship-based preference multi-objective evolutionary algorithm .By stratifying the fitness on the non-dominant individuals set by a neighbour dominant relationship and using the reference point , the algorithm guides the individual population approaching the interested region of the decision maker .According to the comparison experiments with some classical preference multi-objective evolutionary algorithms , the results show that the proposed algorithm can well guide the population closing to the most satisfying areas of the decision makers .