| With the rapid development of high-speed railways,the increasing mileage of China’s railway construction has reached the forefront of the world.In order to ensure the safe and efficient operation of high-speed railways,it is important to study the fault analysis of on-board equipment in Chinese Train Control System(CTCS).In the operation of vehicle-mounted device,corresponding log files are generated.The log files contain a large amount of device operation information that are the record of the vehicle-device failure information.So efficient and accurate extraction of key fault information in log files is of great significance for fault identification and analysis of equipment during high-speed train operation.The breakdowns of on-board equipment are various and uncertain,and the staff relies heavily on personal experience based on the fault information recorded by the DMI.Therefore,the accurate judgment of the breakdown of the on-board equipment needs to be based on the original log file.This topic analyzes and sorts the fault information in log files,combines professional analysis and processing methods for on-board equipment failures,and uses the Matlab simulation platform to build a fault analysis tool,thereby improving the efficiency of staff in handling vehicle equipment failures.The main works of this subject include the following aspects.First,the composition,fault characteristics and types of on-board equipment are introduced,and the initial fault vocabulary is formed by combining the contents of the on-board equipment log files and the written records of professionals.Then,according to the characteristics of the log file,automatic pre-processing of the Log file and the text records of staff is realized,which is suitable for future research.Secondly,the manuscript uses the hidden Markov model-based text information extraction method to process and explore the log files.Before performing key information extraction,marks the pre-processed log files to establish the initial character dictionary.A trained model file is estimated,and then the test samples are labeled and extracted according to the trained model file to achieve the extraction of the fault information in the log file.Finally,based on the Matlab platform,a fault analysis tool is developed to accurately locate the fault information in the Log file and analyze and process the fault.At the same time,it can identify the unrecorded fault information.The manucript analyze the advantages and disadvantages of this tool,and make some prospects for further research. |
