Research On System-level Fault Diagnosis Based On High-speed Railway Operational Scenarios

Railway signal equipment faults and the incomplete fault diagnosis system is the main reason for the railway accidents such as:rear-end collision, overrun and derailment. Although the traditional technology of fault diagnosis has been continuously developing, it still has some drawbacks. The occurrence of the accident 7.23 indicates that fault diagnosis technologies rely on the level of subsystem, equipment or device cannot meet the requirements of safety, a more effective and accurate fault diagnosis system on system-level becomes particularly necessary.The main innovative research results of this paper are as follows:(1) In order to solve the limitation of traditional fault diagnosis technology, this paper proposes a system-level fault diagnosis method based on high-speed railway operational scenarios. In the high-speed railway operation process, the system-level fault diagnosis system applies big data and data mining technology to monitor the real-time data of high-speed railway signal equipment to detect and locate the faults. Compared to conventional fault diagnosis technologies, it makes diagnosis more timely, accurate and effective.(2) Combining the characteristics of big data and data stream, this paper introduces high-speed railway operational scenarios’data stream and its mining technology and proposes a system-level fault diagnosis method and then established the corresponding model. This method acquires high-speed real-time data online by using the technology based on a sliding window. The system fault data set can be acquired through dynamic meshing and outlier detection after the division of signal equipment clusters and preprocessing. Finally, association rules are used to analyze the fault diagnosis.(3) In the process of mining, this paper pretreats the data of high-speed railway operational scenarios’ data stream to remove the data records that are not directly associated with the clusters of signal equipment to reduce the size of mining and enhance the mining efficiency, then replaces the state of each signal device with the corresponding figures to be converted into a form which is suitable for data stream mining.At last, this paper validates the proposed method through the siding pick scene of Jiang-ning Xi station of Ning-an Passenger Dedicated Lines (PDL), the results indicate that the data stream mining is useful for system-level fault diagnosis of high-speed railway operational scenarios’ fault cases, and play a certain role in the overall advantage of high-speed railway signaling system.