摘要
提出了用于求解大规模优化模型的基于网格划分的混合算法。该算法引入了空间划分和收缩的思想,在求解过程中首先应用全局优化算法确定优解信息,其次使用网格划分和合并将解空间快速划分和收缩为多个子空间,然后用局部优化算法在模型的极值点附近搜索,可以很快地收敛到极值点。仿真结果表明该算法在搜索效率、应用范围、解的精确性和鲁棒性上都体现了良好的性能。
A hybrid optimization algorithm based on grid partitioning for solving large-scale optimization is proposed. The algorithm adopts the idea of space partitioning and contracting. Firstly, the global optimization algorithm is used to gain the information of elite solutions. Then the grid partitioning and uniting are organized to divide and contract the solutions space as multi-subspaces. Finally, in order to obtain the extrema the local optimal algorithm is applied in these subspaces. The numerical simulation result shows that the proposed algorithm is robust, effective, and efficient.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2007年第2期312-315,共4页
Systems Engineering and Electronics
基金
山东省自然科学基金(Y2003G01)资助课题
关键词
网格划分
优化
演化算法
grid partitioning
optimization
evolutionary algorithm
