Research On Signaling Infrastructure Maintenance Strategy Based On Big Data Risk Analysis

Safety is always the key topic of the railway system operation.As the core component of the railway operation system,signaling system has been established a more important reputation in keeping the safety and efficiency of the train operation.With the rapid increase of the passengers,much more work pressure has been given to the signaling infrastructure.To find a more efficient maintenance strategy that keep the safety and efficiency of signaling infrastructure is a more serious problem to be solved.With the development of the computer monitoring system of signaling system,the collected monitor data has been given more big-data characters.But the data haven’t been efficiently used.In this thesis,we will focus on the big data analysis technology.The risk level of the signaling infrastructure will be analyzed with the big data processing technology,considering with the data character of the signaling infrastructure monitoring data.The new optimized maintenance strategy will be set according to the risk level.The main contributions of this work are summarized as follows:Firstly,according to the character of the signaling system monitoring data,an optimized data process algorithm is proposed to analyze the massive small files of monitored signaling infrastructure data with the MapReduce calculate structure.The SequenceFiles methodology was introduced to optimize the FP-Growth Algorithm.The inherent defects could be eliminated and the data process efficiency could be optimized with implementing the MapReduce-based SequenceFiles-optimised FP-Growth algorithm.Secondly,a new risk-based maintenance strategy optimization method is proposed.The risk level is generated with the association rules mined from historical collected data and risk level data.The optimized FP-Growth algorithm is used for association rules mining.The real time risk level of infrastructure is calculated with the association rules.The maintenance strategy is decided according to the current scheduled maintenance strategy and risk level.Finally,the big-data-risk-analysis based on maintenance strategy is implemented with the data of point machine.The monitored data and environmental data are used as input to generate the association rules.With the trained data model,risk level could be calculated based on the collected data.The maintenance strategy is optimized with the risk level.The feasibility of the methodology is proved by the case study through analyzing the efficiency of the maintenance strategy.