摘要
考虑到高铁车载设备故障诊断的不确定性和复杂性,本文提出了基于贝叶斯网络的车载设备故障诊断系统。在建立贝叶斯网络结构的过程中,基于充分利用现场数据与先验知识的思想,本文通过融合不同方法(K2算法,MCMC算法和专家知识)得到最优的贝叶斯网络结构。最后,本文进行了实例分析与模型验证,并与KNN算法、BP神经网络算法进行比较,测试结果表明该模型的正确性和有效性。文中的验证数据来自武广高铁车载设备故障追踪表。
Due to the uncertainty and complexity of fault diagnosis of vehicle on-board equipment (VOBE)of high speed railways,the Bayesian network (BN)based fault diagnosis system for VOBE was put forward.In Establishing the BN structure for VOBE,by fully utilizing site data and a priori knewlege,different methods (Algorithm K2,Algorithm MCMC and expert knowledge)were compromised to set up the optimum BN struc-ture.Case study and model verifyication were carried out.Comparison with Algorithm KNN and Algorithm ANN-BP shows that the proposed BN model is correct and effective.The field data are from the fault detection and diagnosis table of VOBE of the Wuhan-Guangzhou High-speed Railway.
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2014年第11期48-53,共6页
Journal of the China Railway Society
基金
中国铁路总公司重点项目(2013X015-B)
轨道交通控制与安全国家重点实验室自主研究课题(RCS2012ZT005)
关键词
贝叶斯网络
故障诊断
高速铁路
车载设备
Bayesian networks
fault diagnosis
high-speed railway
vehicle on-board equipment
作者简介
赵阳(1990-),男,山西运城人,硕士研究生。E-mail:12120304@bjtu.edu.cn
通讯作者:徐田华(1971-),男,山东临沂人,副教授,博士。E-mailthxu@bjtu.edu.cn