摘要
本文讨论上层决策给定的条件下，下层存在多追随者的多目标分层次诱 导决策问题．在线性情况下，此类问题的最优解可在有界多面体区域的某个端点实 现；应用罚函数理论，原决策问题转换为一个在有界多面体区域求连续凸函数最大值 的最优化问题．建议采用的计算方法较为简单，容易实现，而且能够保证求出问题的 全局最优解．
This paper deals with a two-level decision making problem with the presence of multiple followers. In linear case, the optimal solution to this tape of problem is achievable at one of the vertices of the bounded polyhedral region of the feasible decisions. Using penalty function, the two-level problem is transformed to a special optimization problem where a continuous convex function is maximized within the bounded polyhedral region. The transformed problem can be solved efficiently using a cutting plane algorithm and the solution is a globally optimal one.
出处
《运筹学学报》
CSCD
1999年第3期25-34,共10页
Operations Research Transactions
基金
国家教委留学回国人员科研基金
关键词
多目标决策
非凸性规划
分层次诱导决策
最优解
Two-level Optimization
Multi-Criteria Decision Making
Non-Convex Optimization
Global Optimization.
