摘要
如何评价智能优化算法在有限时间内所得解的质量,是计算智能基础研究和工程实践中都亟待解决的问题.受序优化思想启发,针对连续优化问题,提出一种评价智能优化算法解质量的方法.首先利用聚类方法对解记录均匀化分区,然后根据适应度值分布计算对准概率作为解质量评价指标.通过对均匀采样、非均匀采样、粒子群算法和遗传算法的寻优结果进行实验表明了所提出方法的有效性.
In the domain of computational intelligence and its applications, how to evaluate the quality of solutions obtained from intelligent optimization algorithms in finite time is an urgent problem to be solved. For continuous optimization problems, an approach is proposed to evaluate the solution quality of intelligent optimization. Firstly, clustering is employed to partition the solution-record. Then, based on the fitness distribution, an alignment probability is calculated as the quality measure. Experiments are performed on uniformly-distributed search, nonuniformly-distributed search, particle swarm optimization and genetic algorithm, respectively. The simulation results show the effectiveness of the proposed method.
出处
《控制与决策》
EI
CSCD
北大核心
2013年第11期1735-1740,共6页
Control and Decision
基金
国家自然科学基金青年基金项目(61105126)
高等学校博士点专项基金项目(20100201110031)
关键词
智能优化算法
解质量
聚类
序优化
intelligent optimization algorithms
solution quality
clustedng
ordinal optimization
