摘要： 针对传统优化方法在处理带约束的多目标优化问题上的不足进行了分析,将多目标进化算法以及约束支配的概念结合起来,重新定义了种群个体间的支配关系,避免了罚函数法因惩罚系数不合适而出现优化结果为非可行解的情况。并且结合惩罚值改进了选择算子和适应值分配机制,避免出现早熟收敛。同时,采用精英策略,让精英个体参与遗传操作,加快算法收敛速度。通过算例分析可知,将多目标进化算法以及约束支配的概念应用到浮筒配置优化方案是可行的、有效的。

关键词: 多目标进化算法, 精英策略, 约束处理, 浮筒配置

Abstract: Combining the multi-objective evolutionary algorithm and the concept of constrain-domination to solve the multi-objectives problems with constraints, this paper redefines the domination relation among individuals in the population and improved the selection operation to avoid the appearance of the infeasible solutions for the improper penalty parameters. Besides, the selection operator and the method of fitness assignment are improved to avoid premature convergence and the elistist strategy is adopted to accelerate the convergence speed of the algorithm. The numerical example shows that the algorithm is effective to solve the buoys-arranging problem.

中图分类号: 

• TP301