基于递阶优化算法的一类两层规划问题的解法  

Hierarchical optimization algorithm-based solutions to a class of bilevel programming problems

作　　者：郑丕谔[1] 刘国宏[1] 李瑞波[1] 赵玉超[1] 

机构地区：[1]天津大学系统工程研究所,天津300072

出　　处：《系统工程与电子技术》2005年第4期662-665,680,共5页Systems Engineering and Electronics

摘　　要：提出一种基于分解协调的两级递阶结构优化算法来求解两层规划问题。通过设计解耦变量,两层规划问题被分解成若干相互独立的易于在结构的第一级求解的子问题。而结构的第二级是调整解耦变量使各子问题的解得以改善。算法以一种迭代的方式使第一级求得的子问题的解不断协调,最终达到两层规划的解。A decomposition-coordination-based optimization algorithm in a two-level hierarchical structure is proposed to solve a bilevel programming problem. By introducing decoupling variables, or coordinating variables, the programming problem is decomposed into some independent subproblems which are easily solved at Level 1 of the structure. And at Level 2 the coordinating variables are updated to improve the solutions to subproblems. The algorithm is carried out in an iterative fashion in order to continuously coordinate the solutions towards the optimal one to the bilevel programming problem. A case study demonstrates its feasibility and effectiveness.

关 键 词：递阶优化算法 两层规划 分解-协调 拉格朗日乘子法 

分 类 号：C934[经济管理—管理学]