基于贝叶斯网络的不确定环境装备故障推理模型

Applying BN(Bayesian Network) to Establishing a New and Effective Failure Inference Model of Equipment under Uncertainties

针对不确定环境下装备故障传播及推理问题,提出了一种基于贝叶斯网络的故障推理模型,利用网络结构与概率分布有效表达装备中各部件故障状态、关联关系及传播方式。首先将模型中变量按照其对应部件在装备中所处地位及层次的差别进一步分为故障检测变量、故障原因变量与故障模式变量三个子集。其次,依据维修人员在故障推理过程中的思维方式,提出了一套符合故障推理任务的模型网络结构有向边取向规则。然后,分析故障推理模型中变量条件概率分布特点,明确其在不确定性表达及参数简化中的优势。最后,建立平视显示器的故障推理模型实例,结合贝叶斯网络推理能力进行故障预测及诊断分析,验证模型的有效性。

Aim. The introduction of the full paper reviews some papers in the open literature and points out what we believe to be their shortcomings; then, it reviews some other papers on successful BN applications; finally, it proposes what we believe to be a new and effective application mentioned in the tide. Section 1 explains how we established our failure inference model based on the BN ; its core consists of： （ 1 ） our failure inference model uses the network topology and the probability distributions to represent the components, relationships and propagations in the equipment; （2） we divide the variables into failure detection subset, failure cause subset and failure mode subset according to their levels and causalities in the equipment; （3） we put forward the network edge orientation rule based on the maintenance engineers＇ actual failure reasoning processes; （4） we analyze the conditional probability distributions of the variables in the failure inference model to indicate their advantages for uncertainty representa- tions and parameter reductions. Section 2 does the case study of a head-up display failure inference model; the results, given in Tables 4, 5 and 6, and their analysis show preliminarily that our failure inference model based on the BN is effective for equipment failure diagnosis and prediction.