The study of multi-objective decision method based on Bayesian network
The study of multi-objective decision method based on Bayesian network
复制标题
基于贝叶斯网络的多目标决策方法研究
DOI：
10.1109/icieem.2010.5646523
复制
发表时间：
2010
期刊：
影响因子：
--
通讯作者：
S. Si
中科院分区：
文献类型：
--
作者：
Hongyan Dui;Li;Shu;S. Si
在对层次分析法（AHP）、模糊综合评判法和贝叶斯网络进行研究的基础上，采用将指标及其关系转化为贝叶斯网络中的节点和有向边的方法，建立了多目标决策贝叶斯网络模型（MODBN）。首先，利用模糊综合评价中指标体系等级域的模糊关系矩阵计算MODBN中叶子节点的先验概率。其次，根据层次分析法的指标权重计算MODBN节点的条件概率。第三，MODBN不仅可以优选出一个方案并确定其状态，还可以在综合重要性分析的基础上找出关键因素并提出方案的优化措施。最后通过科研项目应用实例与层次分析法的比较，验证了该模型的有效性和实用性。
Based on the study of analytic hierarchy process (AHP), fuzzy comprehensive evaluation and Bayesian network, this paper establishes the multiple objective decision Bayesian network model (MODBN) by use of the method of converting the indexes and their relationships into the nodes and directed edges in Bayesian network. First of all, calculate the prior probabilities of the leaf nodes in MODBN using the fuzzy relation matrix of grade field of index system in fuzzy comprehensive evaluation. Second, calculate the conditional probabilities of the MODBN nodes by the index weights of AHP. Third, MODBN can not only optimally select one scheme and determine its state, but also find out the key factors based on the comprehensive importance analysis and propose the optimization measures for the scheme. At last, the effectiveness and practicality of the model are verified by the comparison of the example of application for research projects and the analytic hierarchy process.