不确定环境下多阶段多目标决策模型  

Multi-stage and multi-object decision making model under uncertainty

作　　者：蔡志强[1,2] 孙树栋[1] 司书宾[1] 王宁[1] 

机构地区：[1]西北工业大学机电学院现代设计与集成制造技术教育部重点实验室,西安710072 [2]巴黎中央理工大学工业工程实验室,沙特乃玛拉布利92290

出　　处：《系统工程理论与实践》2010年第9期1622-1629,共8页Systems Engineering-Theory & Practice

基　　金：国家863/CIMS计划基金(2007AA04Z187;2007AA040701);航空科学基金(2009ZE53052);中国留学基金委中法博士生学院基金;西北工业大学翱翔之星基金(07XE0152);陕西省科技计划(2010K8-11)

摘　　要：在贝叶斯网络的基础上进行扩展,提出了一种不确定多阶段多目标(Uncertainty,multistageand multi-object,UMM)决策模型来求解不确定环境下的多阶段多目标决策问题.首先描述了不确定环境下多目标多阶段决策问题的数学描述,然后给出了决策模型的定义,详述了其构建方法及求解算法.采用此模型求解决策问题时,决策者只需考虑某节点与其父节点之间的依赖关系,降低了思考的复杂程度,适用于大规模的复杂问题求解.实际案例分析表明基于贝叶斯网络的决策模型描述、建模及求解方法对不确定环境下的多阶段多目标决策问题是有效的.A newly defined UMM （uncertainty, multi-stage and multi-object） decision making model is illustrated based on the Bayesian network to deal with the multi-stage, multi-object decision making problem under uncertainty. The mathematic description of this decision problem is put forward at first. The definition of UMM decision model, the modeling methodology and the calculation algorithm are all described in details afterwards. With the UMM model, the decision maker could just focus on the relationships between nodes and their father nodes, which makes the model application more convenient. At last, the procedure for implementing the UMM decision model is demonstrated based on a case study of maintenance scheme decision making. With the case study, it is verified this model＇s effectiveness in solving the multi-stage, multi-object decision making problem under uncertainty.

关 键 词：贝叶斯网络 不确定性 多阶段 多目标 决策模型 

分 类 号：O224[理学—运筹学与控制论]