Research On Urban Rail Transit Signaling System Health Maintenance Platform Based On Big Data

With the rapid development of urban railway system, the transportation pressure is increasing in the railway transit network, and put much more pressure on system maintenance. The current rail transit signaling system has problems such as dispersion maintenance modes, incomprehensive data collecting, imperfect data analysis and backward management methods, which means that traditional maintenance methods and technology management methods cannot meet the needs of modern maintenance support. The health management as a crucial technique of the autonomous complex system safeguards, which can enhance the technique, equipment and management skills of the railway signaling system maintenance support, achieves the function of health condition monitoring and comprehensive maintenance management of the system, and provides support to the condition-based maintenance and comprehensive protection of the whole system.For the realization of data-driven based system health maintenances, big data can not only solve the problem of analyzing massive maintenance data, but also provide comprehensive business structure for the signaling system during the health maintenance process. It can meet the needs of the whole process, which beginning from data collection to data storage, and finally to the whole technique and management needs of data integrated application and information release process. Therefore, the signaling health maintenance platform based on the business ideas of big data is written in this paper. The benefit of this platform is to realize the linkage of the health condition of the signaling system and put a whole process control to the health management system, which can help the management department to utilize the advancement of the rail transit through the platform, and carry out the operating security on a system scale.The paper first combines the application of big data in the health maintenance of the signaling system, and analysis the health maintenance big data and application sight of the knowledge discovery, and use the failure correlation principle as an example to research the algorithm and realization of health maintenance distributed data mining. Based on the analyzation of the platform equipment use and the business needs of health management, the paper is discussed in the following order. First is the level of the platform architecture design, which clears the structural components and functions of the data collection layer, the data storage layer, the comprehensive application layer and the information release layer. After that is the health maintenance business design of the signaling system, and divided the business process and management style into three parts, which clearly ensure the structure and function of each layer.In view of the platform health maintenance of the big data business structure, the paper researches the crucial techniques and realization of each layer. In the data collection layer, it studies the maintenance data sources and collection method of the signaling system, and the data preprocessing and data transmission way. Data storage layer is based on the realization of Hadoop technique. By using the HDFS to complete the storage of the original data, and enhancing the efficiency of small files by merging the sequence files. Furthermore, the function of data access and query can be realized by setting up an HBase file mapping. Comprehensive application layer provides the interface of the big data analysis business of the data center. In the integrated application layer, it realizes the real time health condition monitor of the signaling system and fault alarming; apart from this is the history conditions review and statistical failure analysis. By using work order management and equipment asset management, the system can achieve the comprehensive maintenance business; and by using agency authority management, the system can achieve business process management. In the data release layer, it studies the information release function of the health maintenance platform and terminal access and the realization of the data release procession.In order to complete the software and hardware exploration, the paper verified this platform implementation through the laboratory test environment. First set up Hadoop cluster to experiment the storage function of the health maintenance data, which prove the feasibility of health maintenance data scheme and technical advantages. Following that is a laboratory simulation experiment platform to test the function of health maintenance business. And the results confirm that the expectation of the platform is realized and it can fully meet the health maintenance requirements of the signaling system.