基于贝叶斯网络的人因可靠性评价

提出一种贝叶斯网络的人因可靠性评价(HRABN)方法,其中的每个因子对应于贝叶斯网络中的节点,该方法可对人因可靠性作定量分析和定性分析。在定性分析上,节点的因果关系(HRA中的因子关系)及需要改进的薄弱节点都直观地显示在层次图中;在定量分析方面,对节点因子后验概率的推断通过HRA中的先验信息(包含仿真数据、现场操作及专家知识等)和最新信息得到。如果人因可靠性贝叶斯网络中的每个节点的先验概率分布和后验概率分布都已知,模型的可信性就可通过贝叶斯因子进行定量验证。贝叶斯网络扩展性好,当有新的节点因子需要考虑时,只需要补充对应的节点;笔者的方法也能很好地应用在不同行业的HRA。

Human Reliability Assessment Based on Bayesian Networks

A human reliability assessment method using Bayesian networks is presented,in which each factor in the human reliability assessment corresponds to a node in the Bayesian networks,and could be used in qualitative and quantitative analyses.In the qualitative analysis,the causality of the nodes(the factors in the HRA) and the weak points need to be improved will be shown directly through hierarchical graph.In the quantitative analysis,the posterior probability(the potential factor) is inferred by the prior information(including simulation data,onsite experience data and expertise knowledge) and latest information of HRA.A certain potential human actions could be predicted by mathematical expectation of the node's posterior probability.The confidence of the model of HRABN might be quantitatively analyzed if the prior probability distribution and posterior probability distribution of every node were known.In addition,the flexibility of Bayesian networks is well,only corresponding nodes are added when new factors must be taken into account.The method could be well applied to every aspect in HRA.