Abstract: In this paper，Max-Min fitness function and penalty function are combined together，and a practical and effective particle swarm optimization algorithm is proposed to solve multi-objectives constrained optimization problems.Non-inferior solutions are replaced according to the idea of cluster and compare.The method of selecting globally optimal solution from non-inferior solutions in turn is adopted instead of the ancient method.The experimental results show that the modified MaxMin-PSO algorithm converges more quickly and efficiently to Pareto solutions and achieve a well distribution.It also restrains the radiation of low dimension multi-objectives constrained functions.

Key words: particle swarm algorithm, MaxMin function, turn list, punish function

摘要： 将最大最小化适应度函数与罚函数相结合，提出了一种实用有效求解多目标约束优化问题的粒子群算法。采用归类和比较的思想进行替换非劣解；改变以往全局最优值的选取方法，而采用轮序方式从非劣解中获取。实验证明改进的MaxMin-PSO算法能更加有效的逼近Pareto解，收敛速度更快，分布更均匀，且能很好的抑制低维多目标约束问题的发散现象。

关键词: 粒子群算法, 最大最小适应函数, 轮序, 罚函数