基于故障树与贝叶斯网络的高速动车电机组早期故障率估计

Early failure rate estimation of high-speed EMU motor set based on fault tree and Bayesian network

复兴号动车电机组包括牵引电机、冷却风机及相关传感器,是动车组的动力单元.针对其早期故障分布特点不明确及早期故障率估计不准确的问题,利用动车组实际运用的故障数据,从故障整体、故障类型和高频故障致因三方面分析故障分布特点,得到电机组4类早期常见故障和27种故障致因.建立电机组故障树模型,通过定性分析和定量计算,得到早期各个事件故障率.运用贝叶斯网络对各故障的后验概率进行估计.结果表明:可以依据故障树结果划分等级制定维修计划,依据贝叶斯网络确定故障排查顺序.

The motor set of Fuxing EMU includes traction motors,cooling fans and related sensors,which is the power unit of the EMU.Aiming at the problems of unclear early fault distribution characteristics and inaccurate early fault rate estimation,this paper analyzes the fault distribution characteristics based on the fault data of EMU obtained in actual operation from three aspects:overall fault,fault types and high-frequency fault causes.And four types of early common faults and 27 fault causes of the motor set are obtained.A fault tree model of the motor set is established,getting the early failure rate of each event through qualitative analysis and quantitative calculation.Finally,Bayesian network is used to estimate the posterior probability of each failure.The results show that the maintenance plan can be made according to the classification of the fault tree results,and the troubleshooting sequence can be determined based on the Bayesian network.