复杂不确定系统可靠性分析的贝叶斯网络方法

针对结构复杂、实验样本有限、可靠性数据不足等因素导致的复杂多态系统可靠性分析不确定性问题,将模糊数学和灰色系统理论引入贝叶斯网络模型中,提出一种基于不确定隶属度函数和区间特征量的复杂不确定系统可靠性分析方法。用含变量的隶属度函数来表征组件故障状态,有效地避免了人为主观因素对隶属度函数选择的影响,解决了组件和系统的故障状态难以准确界定的问题;用区间灰数描述条件概率表中的确定值,表达组件和系统间不确定的故障逻辑关系;构建出系统可靠性特征量的参数规划模型,以区间的形式表示系统可靠性特征量。将所提方法应用到卫星推进系统的可靠性分析中,研究结果表明,该方法能够有效分析模糊不确定条件下的系统可靠性和组件重要度,且计算量可控,是一种有效的复杂不确定系统可靠性分析方法。

Bayesian Network Method for Reliability Analysis of Complex Uncertainty Systems

The uncertainty in reliability analysis of complex multi-state system may be due to the complexity of system structures, the limited test samples, and the insufficient reliability data. A new reliability analysis method for complex uncertainty system based on non-deterministic membership functions and interval characteristic quantities is proposed by introducing the fuzzy mathematics and grey system theory into the Bayesian network model. The influences of human subjective factors on the selection of membership function can be effectively avoided and the failure states of components and systems can be defined accurately by using the variable membership function to describe the failure states of component and system. The uncertain failure logic relationship between component and system can be effectively expressed by substituting the exact value in the conditional probability table with the interval grey number. A parameter planning model of the system reliability characteristic quantity is constructed, and the system reliability characteristic quantity is expressed in the form of an interval. The proposed method is applied in the reliability analysis of the satellite propulsion system. The results show that the method can be used to analyze the system reliability and component importance under fuzzy uncertain conditions, and the calculation amount is controllable. It is an effective reliability analysis method for complex uncertainty systems. Key