摘要
针对多目标优化问题,提出了一种新的基于实数编码的线性进化算法。新算法将约束优化问题的高维搜索空间通过线性变换映射到二维空间,在二维空间中探索原优化问题的解,并构造出一种线性适应度函数,重新设计了一种基于密度函数的交叉算子。对二组典型优化问题的测试表明,本算法是可行和有效的,解集分布的均匀性与多样性均较理想。
A new Multi-objective Linear Evolutionary Algorithm (MOLEA) based on real-coded for constrained multiobjective optimization was proposed. Search space of constrained dominance problems with high dimensions was compressed into two dimension in the LEA, which contains two main points. Firstly, gave a linear fitness function in two dimension space. Secondly, gave a erossover operator based on density function. In our tests, a few benchmark multi-objective optimization problem which was divided into two groups were taken to test this algorithm. The numerical experiments show that proposed approach is feasible and effective, and provide good performance in terms of uniformity and diversity of solutions.
出处
《计算机科学》
CSCD
北大核心
2009年第4期235-238,共4页
Computer Science
基金
国家自然科学基金资助项目(60472060)
江苏省计算机信息处理技术重点实验室开放课题基金资助项目(KJS0601)
江苏省“青蓝工程”资助
关键词
多目标优化
进化算法
PARETO最优解
线性函数
Multi-objective optimization, Evolutionary algorithm, Pareto optimal solutions, Linear function