基于GERT网络的非等概率共因失效系统故障间隔期预测

MTBF Prediction of Non-equiprobable Common Cause Failure Systems Based on GERT Network

利用泊松过程描述了系统共因失效的发生规律,并针对冲击应力下各部件失效概率不同的假设,以状态转移概率和冲击应力间隔时间为GERT网络的参数,构建了非等概率共因失效下的故障间隔期预测模型,并利用流图增益矩阵进行传递函数的快速求解。按照系统最小割集划分故障模式,将系统分解成若干个GERT子网络,并分析各自故障间隔期,以间隔时间最短的最小割集作为决定系统平均无故障时长的关键割集。与已有文献中的方法进行对比,详细讨论了本文所提方法计算非等概率共因失效系统故障间隔期的优势。最后的仿真结果表明,本文提出的模型能够正确预测系统的平均无故障时间。

This paper employed Poisson process to describe the occurrence of common cause failure.Besides,it set transition probabilities and Time Between Shocks as the parameters to build the MTBF (Mean Time Between Failure)prediction model on the assumption that component faults occurred with variable probabilities,and the gain matrix of flow graph was used to solve the model.Meanwhile,the system was decomposed into several GERT subnetworks by means of finding all the minimal cut sets.Each subnetwork was analyzed and the one with the shortest MTBF determined the entire system's MTBF.Moreover,compared with an existing method,the advantage of the proposed method for calculating the MTBF of non-equiprobable common cause failure systems is discussed in detail.Finally,the simulation results demonstrate that the proposed model is able to predict the MTBF of system accurately.