证据网络模型及其推理算法

Evidential network model and reasoning approach

本文在证据理论与图模型研究的基础上,针对不确定性与关联性相结合的系统建模问题,提出并建立了证据网络模型,定义了证据网络的参数,总结了证据网络的建模特点;分析了证据网络模型的推理问题,建立了其推理策略和推理过程,推导出其正向推理算法和反向推理算法,并通过示例进行了计算验证;证据网络模型将为管理决策分析、系统工程管理等领域的问题分析和求解提供一种有效的技术方法支持.

The evidential network model is proposed, on the basis of the Dempster-Shafer theory of evidence and the graph theory, for management problems with uncertainty and complex correlation. The definition and parameter model of evidential network is presented before the advantages of evidential network are concluded. The reasoning problem, reasoning framework and reasoning process are proposed. Then the forward and backward reasoning algorithms are deduced. Finally, a number example is explored to demonstrate the algorithm and process. The evidential network model will be provided as an efficient method for analyzing and resolving the problems in the research fields of decision analysis and system engineering.