基于微分进化求解多目标优化问题中的退化现象

Deterioration in solving multi-objective optimization problems base on differential evolution

基于微分进化(DE)的多目标进化算法(MOEA)在求解过程中存在着退化现象,导致算法的收敛性无法保证,同时也降低了求解的效率。针对这一问题,分析了算法中存在的两种退化现象,提出了针对两种退化现象相应的解决办法,最后给出了一种新的基于DE的MOEA。新算法克服了已有算法中存在的退化现象,保证了算法的收敛性和解的多样性,有效地提高了算法的效率,通过数值实验验证了新算法的可行性和有效性。

It was found that there exist deteriorations in Optimization Problems （MOP） base on Differential Evolution he process of sol （DE）. The deteri the convergence of the multi-objective optimization algorithm is no 1 reduces the efficiency of the solution. To overcome this problem, two optimization algorithm were analyzed, and corresponding methods to were proposed. A new Multi-objective Optimization Evolution Algo which guarantees the convergence significantly improved. Numerical e ving Multi-objective oration onger guaranteed, leads to that and it also deterioration phenomena in the resolve the two deteriorations rithm （MOEA） is proposed, and the diversity. The efficiency xperiments validate the effectiveness o of the new algorithm f the new algorithm.