高速铁路信号系统异构数据融合和智能维护决策
Heterogeneous Data Fusion and Intelligent Maintenance Decision for High Speed Railway Signaling Systems

针对高速铁路智能维护决策中的信息多源异构问题,提出了高铁信号系统异构数据融合和智能维护决策架构。通过本地数据库资源描述框架(resource description framework (schema),RDF(S))到全局RDF(S)的转换和基于RDF(S)图的本体合并,实现了多源异构信息的融合。利用适合缺失数据的结构期望最大化(SEM)算法,结合专家知识,构建了高铁信号系统的智能维护贝叶斯网络(BN)决策模型。最后,利用高铁武广线2011～2012年的监控数据,分别对基于RDF(S)图的全局RDF(S)合并算法性能和故障诊断结果的准确性进行了分析对比,实验结果表明所提出的本体融合算法具有多项式级的计算复杂度,同时融合专家知识和SEM算法的智能维护BN决策模型的一级故障诊断准确率为92.4％。因此,所提出的异构数据融合和智能维护架构可以有效提高高铁信号系统维护决策的准确性和有效性。
A framework of integrated heterogeneous data and intelligent maintenance decision was proposed aiming at the multi？？source heterogeneous data in intelligent maintenance decision for railway signaling systems. By means of transformation and fusion from local resource description framework (schema)(RDF(S)) to global RDF(S), the fusion of heterogeneous data was realized. In addition, a Bayesian net (BN) based intelligent maintenance decision model was constructed by combing the structural expectation maximum (SEM) algorithm for missing data with the expert knowledge. The correctness and efficiency of the proposed framework and the RDF(S) fusion algorithm were verified by the maintenance data from Wuhan？？Guangzhou high？？speed railway signaling systems in 2011？？2012. The experimental results show that the computational complexity of the proposed ontology fusion algorithm is polynomial, and the average accuracy of the fault diagnosis for the first level reaches 92.4%. Therefore, the proposed framework may improve the accuracy and efficiency of the intelligent maintenance decision of high？？speed railway systems