The Analysis And Design Method Of Big Data-Driven Railway Cyber-Physical System

In 2015 China’s high-speed operation railway has been more than 18,000 kilometers and the amount of high-speed trains has occupied more than 50 percents of the total number of trains. Speed and Safety is the critical of high-speed train. With the widespread use of Internet of Things and sensors(information of Train or Track states capturing), the railway signal system produces vast amounts data every day. So there is a big chance in intelligent railway(intelligent train and intelligent schedulling, etc.) because there is large hidden value of data in big data. But the vast amouts of data produced from mass sensors brings new challenges:the heterogeneous data, big data storage, data sharing and big data process.Cyber-physical sytem is a system which is a comlex embeded system composed with physical system, information system and control logic deeply. It can perceive physical environment, control dymically and provide information services. At the meantime, it is a real-time, safe-critical and heterogeneous system. As an example, there are many cyber-physical systems in railway systems and the most typical one is the train operation control system. Because of the complexity and large-scalibility of railway cyber-physical system, there are many dynamic and uncertain factors. So, how to analyze and design railway cyber-physical system to know the interaction between physical world and information world, satety, realiability, realtime is a important technology for future railway operation under the background of large-scale application of sensors.This paper combines the big data and cyber-physical system and focus on Big Data and Hybrid system. Firstly, it analyzes the Big Data features of the railway.Then it analizes the intelligent requirements, data structure and technology architecture of railway. As a result, the analysis of the system define the research object. Next, this paper proposes an cyber-physical architecture based on Cloud Computing and analyze the sevice types and heirarchical orchestration of services on the base of railway big data analysis above. The challenges brought from big data driven railway system need to change the traditional railway information system and solve the isolation problems. The Cloud Computing technology is a good method to settle the heterogeneous problems, big data storage problems and information sharing problems. There is one another critical technology we need. Real-time bid data processing is important for railway, so this paper proposes a real-time stream processing method based on Storm for railway big data. As for the "Physical" part in railway cyber-physical system, there is a modeling method combined with Simulink and AADL. To map Simulink models to AADL models will take advantage of the both mostly. It proposes a method to extend the AADL ability to model hybrid behaviors and this make up for the shortage of alone Simulink’s simulation. Finally, this paper studys hybrid behaviors of the critical scene-Move Authorization of train running and the physical system, software system, spatial-temporal behavior modeling of ATP system.