【摘要】 考虑储备集更新策略在多目标粒子群优化算法中的关键作用,提出一种两阶段储备集更新策略。第一阶段,利用自适应储备集更新策略,以保证解的分布性及延展性;第二阶段,利用基于ε占优关系的储备集更新思想,以避免解的退化现象,保证解的收敛性。提出一种新的储备集候选点选取机制以避免相似个体频繁进出储备集;通过优化典型多目标优化问题验证两阶段储备集更新策略及储备集候选点选取机制的有效性。更多还原

【Abstract】 Considering the importance of the archive updating strategy for multi-objective particle swarm optimization,a two-phase archive updating strategy was proposed.In the first phase,the adaptive archiving strategy is used to guarantee good distribution and extension of the solutions.In the second phase,the idea of the archive updating strategy based on ε-dominance relationship is used to preclude the solutions’ deterioration and to guarantee the solutions’ convergence.A new selecting strategy of the archive’s candidates is also proposed in order to avoid the similar individuals entering in and out the archive frequently.Finally,it is shown from experimental results on complicated benchmark multi-objective functions optimization that the two-phase archive updating strategy and the selecting strategy of the archive’s candidates are efficient.更多还原