Abstract: A particle swarm algorithm based on mutation operator is put forward for getting better Pareto solution sets in the field of multi-objective optimization problem. The grey correlation degree theory, which is introduced by variance, is applied to the algorithm for distinguishing the sequences with equal means but sensible difference of object position correlation coefficients. The particle swarm algorithm is controlled by this mutation strategy. Hence, the local convergence phenomenon is prevented during solving multi-objective problem with grey particle swarm algorithm. The algorithm’s performance is tested by four groups of different types of benchmark functions. It shows that the algorithm can not only convergence to the Pareto optimal solution sets  very well but also efficiently avoid falling into the local optimal solution too early.

Key words: multi-objective, mutation operator, grey relational degree, Particle Swarm algorithm, optimization

摘要： 针对如何在多目标优化过程中求解更好的Pareto解集，提出一种基于变异算子的灰色粒子群算法。该算法将灰色关联度应用于粒子群算法，且将方差的概念引入灰色关联度，以区分那些点关联系数有显著差异而其均值相等的各组序列。以此作为变异策略来控制粒子群算法，以避免现有灰色粒子群算法在求解多目标问题时所出现的局部收敛现象。通过四组不同类型的基准函数测试算法性能，测试结果表明该算法能很好地收敛到Pareto最优解集并有效避免了过早陷入局部最优解。

关键词: 多目标, 变异算子, 灰色关联度, 粒子群算法, 优化