Standardization Of Big Data Of High-speed Railway Signal System And Maintenance Decision Support

With the rapid development of high-speed railway in our country, the high-speed railway signal system has produced vast amounts of operation and maintenance data. High-speed railway signal system’s operation and maintenance data has the typical big data characteristics, such as volume, varity and velocity. To realize the analysis and data mining of high-speed railway signal system’s big data, it needs to be standardized and normalized.Among the methods of data normalization, the semantic web, which is through introducing the concept of ontology from philosophy domain, can effectively achieve the semantic information sharing between computers. The semantic web architecture which is put forward by the World Wide Web Consortium (W3C), mainly includes three parts:one is data query which is based on the Resource Description Framework of SPARQL; the second is test of reasoning consistency which is based on Web Ontology Language,(OWL), the third is reasoning of OWL Ontology which is based on the Semantic Web Rule Language (SWRL).In this paper, to deal with the multi-source and heterogeneous problems of high-speed railway signal system’s operation and maintenance data, in reference to the European railway operation and maintenance data processing experience, we propose the high-speed railway signal system heterogeneous data integration and intelligent maintenance decision-making structure on the basis of using semantic web technology and combining with the characteristics of high-speed railway operation and maintenance data in China. To verify the correctness of the structure, this paper uses the wuhan-guangzhou high-speed railway operation and maintenance data from2011to2012of CRH-380A vehicle-mounted equipment. In this paper, the main research contents and research achievements include the following three points:(1) A Hadoop-based multi-source heterogeneous data integration and intelligent maintenance decision support unified architecture for tackling data normalization and to provide intelligent maintenance decisions is proposed which can provide some support and guidance for deep research.(2) MapReduce based local database RDF (S) to the global RDF (S) conversion. The conversion process ontology merging algorithm uses Hoare logic of RDF (S) figure to achieve a fusion of multi-source heterogeneous information. And this paper builds the Onotlogy model of CRH380A vehicle onboard equipment’s operation and maintenance data, which is part of high-speed railway signal system, via ontology modeling tool Protege4.3.(3) For the processed standardization data, using the classic ID3Decision Tree algorithm combining with the knowledge of experts, this paper builds a high-speed railway signal system of intelligent maintenance Decision Tree (DT) model. This paper also do some experiments to verify the merging algorithm’s performance, based on RDF (S) hoare logic diagram to the global RDF (S) fusion, and the accuracy of fault diagnosis the respectively. The experimental results show that the presented fusion algorithm of ontology with polynomial computational complexity, and fusion of expert knowledge of intelligent maintenance decision-making DT model’s fault diagnosis accuracy rate is95.23%.