A Data-driven Approach For High-Speed Railway Train Regulation

China has built a high-speed railway network with the longest mileage and the largest transport density in the world,and the complexity of the operation organization has put tremendous pressure on the high-speed railway dispatching.Train regulation is the core of high-speed railway operation organization,which is of great significance for ensuring the safe and punctual of high-speed trains and improving the quality of passenger transportation services.However,the high-speed train regulation highly depends on the dispatcher’s own experience currently,and the automation degree is low.Thanks to the development of big data technology and artificial intelligence,data-driven methods have been widely used in solving complex problems in engineering applications and scientific research.How to use data-driven methods to improve the effectiveness of train regulation decision-making is becoming an important topic in the research of railway transportation organization optimization.In view of this,based on the real-world train operation records,this paper established a decision-making method for high-speed train regulation.The research contents including:(1)Introduced the basic concepts of high-speed railway train regulation.Classified high-speed railway operation interference,and summarized the corresponding emergency treatment;the relevant theories of high-speed train operation adjustment were summarized,and the common operation adjustment methods and their applicability was analyzed.Based on the analysis of the basis of high-speed train operation adjustment,train operation adjustment process was sorted out in detail to provide theoretical basis for subsequent research.(2)Based on the real-world train operation records,the descriptive statistical results of relevant indicators were visualized from the dimensions of stations and sections,including:initial delay distribution,delay distribution,buffer time distribution and delay recovery distribution.From the macro perspective,the regular of the train delay recovery was shown.(3)From the micro perspective,the delay recovery strategy pre-selection of high-speed train was studied.Based on the analysis of the influencing factors of train operation and machine learning method,with seven factors,which are the arrival time,delay,the scheduled time,the minimum time,the historical average time,the previous train operation constraint,and the scheduled time constraint,the prediction models of delayed train operation and dwell time in the case of delayed recovery were constructed in this paper.The model verification showed that the prediction model has high prediction accuracy.(4)On the basis of the previous research,a practical decision-making method of highspeed train operation adjustment was developed.Described the method design requirements,goals,functions and application process in detail;designed the representing method of modeling basic data;the train arrival and departure time adjustment model was established to make the dwell time adjustment and the operation time adjustment decision;the dynamic adjustment method of train operation sequence was proposed to realize the automatic formulation of train operation sequence adjustment decision.(5)A case of representative actual train operation was studied through the method proposed in this paper,which verified the effectiveness of the model and method in practical application.