Abstract：
Generating feasible solution and selecting valuable solution are the most important issues when dealing with complicated multi-objective problems. Focusing on these issues, the mechanism of multi-objective problem is analyzed by evolutionary history and environmental information. Hierarchical decision based on rank fitness of distance correlation is proposed to guide the evolutionary operator. Heuristic learning by dynamic evolutionary is introduced to deal with static optimization problem. History information acquired from solution landscape is used to achieve a comprehensive search on feasible region. Based on these improvement, multi-objective evolutionary algorithm based on hierarchical decision, heuristic learning and historical environment (MOEA3H) is proposed. The proposed algorithm performs best on 10 and 14 of 19 test problems on IGD and Hvpervolume, respectively.
摘要：
生成可行解和选择有价值的解是处理复杂多目标问题时最重要的问题。围绕这些问题,通过进化历史和环境信息分析了多目标问题的机理。提出了基于距离相关秩适应度的分层决策来指导进化算子。引入动态进化的启发式学习来处理静态优化问题。从解决方案景观中获取的历史信息用于实现对可行区域的全面搜索。基于这些改进,基于分层决策的多目标进化算法,提出了启发式学习和历史环境(MOEA3H)。所提出的算法在IGD和Hvpervolume的19个测试问题中的10个和14个上表现最佳，分别。
