摘要： 基于生物免疫系统的机理及功能，提出一种动态多目标免疫算法。利用抗体的被控度及浓度设计抗体的亲和力。用环境记忆池保存优秀抗体，并依抗体浓度更新。记忆细胞参与相似或相同环境初始抗体群的生成。借助动态多目标测试问题，与同类算法仿真比较，结果表明，该算法较其他算法表现出更好的性能，能快速跟踪动态Pareto面且分布均匀，具有较强的求解实际动态问题的能力。

关键词: 动态环境, 多目标优化, 投资组合, 免疫算法, Pareto面, 环境跟踪

Abstract: Dynamic multi-objective immune optimization algorithm(DMOAIS), which is based on the function of biological immune system , is proposed to solve dynamic multi-objective problems. The affinity of antibody is designed by the strength and crowding distance of antibody. The environment memory pool that is used to saving enlist antibodies is designed for strengthening the diversity of population. Memory cells are participated in the evolution of the similar or the same environment initial population. DMOAIS is compared against other algorithms to solve dynamic multi-objective problems. Numerical experiments illustrate that DMOAIS is promising and competitive to the compared algorithms in solving dynamic multi-objective optimization problems, tracking rapidly dynamic Pareto surface, and showing a powerful exploitation capacity for real-word dynamic multi-objective optimization problems.

Key words: dynamic environment, multi-objective optimization, portfolio, immune algorithm, Pareto surface, environmental tracking

中图分类号: 

• TP301.6