【摘要】 根据可靠性数学理论,从零件失效物理模型应力-强度干涉模型出发,把零件失效概率看作是服从某种分布的随机变量,推导了冗余系统共因失效概率预测模型的数学表达式.通过Monte Carlo仿真法和神经网络技术,得到了该随机变量的分布类型及分布参数.该模型可以根据系统的有限失效数据预测系统任意阶失效概率,能够弥补传统共因失效模型的信息遗漏问题.最后给出实例,并与传统的BFR模型对比,结果表明,该方法比BFR模型更能精确地与实际吻合.更多还原

【Abstract】 According to the mathematic theory of reliability and the physical model of components failure,i.e.,the stress-strength interference model,components failure probability is regarded as a random variable complying with a certain distribution.The mathematical expression of a prediction model of the probability of common cause failure in redundant system is thus derived.Using the Monte Carlo simulation method and neural network technique,the distribution type and parameters of the random variable are obtained.Based on the limited failure data,the model can predict the failure probability of arbitrary order for a redundant system,and it can make up for what have been omitted by the traditional models for common cause failure.A typical example is given to illustrate the application of the model with the calculated results compared to those resulting from BFR model.The results show that the approach proposed is more accurate.更多还原