求解约束非线性优化问题的群体复合形进化算法  

Multi-Complex Evolution Algorithm for Constraint Nonlinear Optimization Problems

作　　者：王建群[1] 卢志华[1] 哈布哈琪[1] 

机构地区：[1]河海大学水文水资源及环境学院,江苏南京210098

出　　处：《河海大学学报（自然科学版）》2001年第3期46-50,共5页Journal of Hohai University(Natural Sciences)

摘　　要：分析了SCE UA算法的特性 ,指出该算法仍存在着一些缺陷 .例如 ( 1)SCE UA算法的全局最优性依赖于随机选取的初始点集的多样性 ,若初始点集选取不当 ,搜索进化就会早熟而陷入局部最优解 ;( 2 )SCE UA算法求解具有区间约束的非线性约束优化问题较有效 ,但对于一般的不等式约束非线性优化问题其求解效率有待于进一步提高 .提出了群体复合形进化算法 ,能充分利用目标函数值的信息 ,优化搜索过程具有较强的方向性和目标性 ,收敛速度较快 ,且是全局优化算法 ,能有效地求解不等式约束非线性优化问题 .The essential concepts and the characters of the SCE UA method are reviewed, and some limitations of the SCE UA method are pointed out, such as (1) the global optimum of the method depends on the variety of the initial point sets, if the initial point sets can not be selected properly, the global optimum can not be obtained,(2) the general inequality constraint nonlinear optimization problems can not be solved effectively. In this paper, a more effective method called Multi Complex Evolution Algorithm for Constraint Nonlinear Optimization Problems is presented. This new method can use the information of the objective function to search for the optimum objectively. This new method is a global optimization method,which can be used for general inequality constraint nonlinear optimization problems efficiently and effectively.

关 键 词：非线性优化 进化算法 复合形算法 全局优化算法 

分 类 号：O221.2[理学—运筹学与控制论]