视情维修条件下的多状态控制单元可用性建模与分析

摘要

针对传统可靠性分析方法难以描述系统动态特性的问题,提出了一种基于多状态动态贝叶斯网络的视情维修可用性建模方法.在定义多状态元件的基础上,构建了基于Markov模型的视情维修状态转移模型,并引入了吸收状态.通过构建多状态动态贝叶斯网络模型,确定了无维修、完全维修、不完全维修、视情维修和吸收状态下的状态转移关系,并根据串联和并联逻辑关系对条件概率赋值.以某控制单元为例,构建动态故障树模型、动态贝叶斯网络模型,确定控制单元与元件在不同维修方式下的可用度变化规律,通过重要度分析发现了可靠性设计中的薄弱环节.仿真分析表明:视情维修的引入使得控制单元与元件相对于完全维修与不完全维修具有更高的可用度;吸收状态的引入可以预测退化状态可修元件的可用度变化趋势,为维修换件提供理论指导.%A modeling method of control units under condition-based maintenance (CBM) based on multi-state dynamic Bayesian network (DBN) is proposed for describing the dynamic characteristics of system.After defining multi-state components,a state-transition model based on Markov model is established in considering CBM,and absorbing state is introduced.A multi-state DBN model is established to determine the state-transition relationships with non-repair,perfect repair,imperfect repair,CBM and absorbing state.And the conditional probability value is obtained according to the logic relationship of series system and parallel system.A control unit is taken for example.The dynamic fault tree (DFT) and DBN model are established to determine the reliability curves of control unit and its components under different repair modes.And the weak links of reliability design are found through importance analysis.The simulated results show that the control unit under CBM has higher availability than the perfectly and imperfectly repaired control units.The availability of repairable components with degradation states can be predicted,and a maintenance policy can be made accordingly.