基于支配强度的NSGA2改进算法  

Improved NSGA2 Algorithm Based on Dominant Strength

作　　者：赖文星 邓忠民[1] LAI Wen-xing;DENG Zhong-min;School of Astronautics;Beihang University;

机构地区：[1]北京航空航天大学宇航学院,北京100191

出　　处：《计算机科学》2018年第6期187-192,共6页Computer Science

基　　金：国家自然科学基金(10972019)资助

摘　　要：NSGA2是一种简单、高效且被广泛使用的多目标进化算法(Multi-objective Evolutionary Algorithm,MoEA),但在求解实际工程领域中的高维、复杂非线性多目标优化问题(Multi-objective Optimization Problems,MOP)时,存在无法有效识别伪非支配解、计算效率低、解集收敛性和分布性较差等设计缺陷。对此,文中提出一种基于支配强度的NSGA2改进算法(INSGA2-DS)。新算法采用快速支配强度排序法构造非支配集,引入了考虑方差的拥挤距离公式,并通过自适应精英保留策略动态调整精英保留规模。基于标准测试函数的仿真实验表明,INSGA2-DS算法较好地改善了NSGA2算法的收敛性和分布性。NSGA2 algorithm is a simple,efficient and widely used multi-objective evolutionary algorithm.However,when solving high-dimensional and complex nonlinear multi-objective optimization problems in practical engineering field,NSGA2 has some obvious design defects,such as ineffective identification of pseudo non-dominated solutions,low computational efficiency,poor convergence and distribution.In order to remedy the above drawbacks,this paper proposed an improved NSGA2 algorithm based on dominant strength（INSGA2-DS）.INSGA2-DS uses the fast dominant strength sorting method to construct non-dominated set,introduces a new crowding distance with considering variance to improve the distribution of solution sets,and adopts the adaptive elitist retention strategy to adjust elitist retention scale in evolutionary process automatically.The experimental results of INSGA2-DS and NSGA2 with standard test functions show that INSGA2-DS algorithm can improve the convergence and distribution of NSGA2 algorithm effectively.

关 键 词：NSGA2 多目标进化算法 多目标优化问题 支配强度 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]