摘要
目前的多目标优化进化算法在高维的决策空间中收敛性能不佳,针对这个问题,提出了基于基因表达式编程的多目标优化算法GEPMO,主要工作包括:提出了新的个体编码方案,分离了值基因和位置基因;设计了新的算子;分析了GEPMO的编码空间;提出了GEPMO的框架。在标准测试函数上的实验结果表明了新算法的有效性,在高维决策空间中GEPMO能够覆盖SPEA算法的结果集87.5%,但SPEA覆盖GEPMO仅为5%。
Traditional Multiobjective Optimization Algorithms are inefficiency in high dimensional decision space. To solve this problem, a novel method named Multiobjective Optimization based on Gene Expression Programming (GEPMO) was proposed. The main contributions of this study include: proposing a new coding method for chromosome, designing some operators for the new coding, analyzing the size of coding space, and presenting the framework of GEPMO. Extensive experiment results on enhanced standard test functions showed that, GEPMO is feasible and effective. In high dimensional decision space, the result set of SPEA is covered by GEPMO at least 87.5% , inversely at most 5 %.
出处
《四川大学学报(工程科学版)》
EI
CAS
CSCD
北大核心
2007年第4期124-129,共6页
Journal of Sichuan University (Engineering Science Edition)
基金
国家自然科学基金资助项目(60473071)
四川省教育厅资助科研项目(2006B067)
关键词
多目标优化
进化算法
高维
基因表达式编程
Multiobjective Optimization
Evolutionary Algorithm
high dimension
Gene Expression Programming