均衡分布性与收敛性的协同进化多目标优化算法

为了进一步提升多目标进化算法(MOEAs)的收敛速度和解集分布性,针对变量无关问题,借助合作型协同进化模型,提出一种均衡分布性与收敛性的协同进化多目标优化算法(CMOA-BDC). CMOA-BDC 首先设置一个精英集合,采用支配关系从进化种群与精英集合中选择首层,并用拥挤距离保持其分布性;然后运用聚类将首层分类,并建立相应概率模型;最后通过模拟退火组合分布估计与遗传进化,达到协同进化.通过与经典 MOEAs 比较的结果表明, CMOA-BDC 获得的解集具有更好的收敛性和分布性.

Co-evolutionary multi-objective optimization algorithm with balanced diversity and convergence

To further improve the diversity and convergence rate of the existed multi-objective evolutionary algorithms, a
co-evolutionary multi-objective optimization algorithm with balanced diversity and convergence(CMOA-BDC) is proposed
specific to the dependency-free multi-objective optimization problems through integrating the cooperative co-evolutionary
model. Firstly, CMOA-BDC sets an elitism set, employs the simple dominant relationship to search the first non-dominant
layer in the evolutionary population and the elitism set, and adopts crowding distance to keep the diversity of the first nondominant
layer. Then cluster analysis is used to divide the first non-dominant layer into multiple class, and the probability
model is established. Finally, a co-evolutionary method is realized by using simulated annealing to integrate the estimation of
distribution and genetic evolution. In comparison with the classical MOEAS, the experimental results show that the algorithm
has better outcomes in both convergence and diversity.