【摘要】 相关失效的存在大大降低冗余系统的安全作用,在工程实际中必须加以考虑。根据可靠性数学理论及零件失效机理分析,把零件失效概率看作是基于应力的条件概率,推导系统相关失效概率的数学表达式。通过MonteCarlo法仿真,得到零件条件失效概率的分布类型;并从已知的低阶失效数据中提取相关失效信息,建立神经网络模型,得到其分布参数。利用该模型可以预测系统中的任意阶相关失效概率,也可以用来预测组成零件相同、环境相同但不同大小的其它系统的相关失效概率。例示其应用方法并检验其预测能力,结果证明该方法准确可靠。更多还原

【Abstract】 Dependent failure can significantly reduce the redundant system reliability,so it has been widely paid attention to in the engineering practice.According to the reliability theory and analysis of failure mechanism on the components,a component’s failure probability was regard to be the conditional failure probability with its stress,so that the mathematical expression for system dependent failure probability was given.Using Monte Carlo simulation,distributed type of conditional failure probability of the components was obtained;Extracting dependent failure information from given low-fold failure data and establishing the neural network model,its distributed parameters were obtained.The model can predict any multiplicity dependent failure probability,and also be applicable to other system with same components and environment and different sizes.An example is provided to illustrate the application,and its predictability is tested,which shows that the approach is accurate and feasible.更多还原