Delay Trend Analysis And Adjustment Of Urban Train Operation Based On Field Data

With the development of urban rail transit in China,the subway has gradually become the predominant travelling choice of urban residents,due to its safe,efficient and punctual characteristics.With the increase of passengers and the decrease of train departure interval,one train delay will bring the delay propagation and thus affect passengers’ experience.The worst case is that it may even cause railway paralysis.Therefore,it is an urgent problem to adjust the train scheduling within limited time to reduce the impact of the train delay.Based on the field data from one in-service urban railway,this thesis proposes a method to analyze the trend of train delay and adjust the delayed train,to improve both train operation efficiency and quality of service.The main contents of the thesis are listed as follows.Firstly,the domestic and overseas research frontiers about train delay prediction and adjustment are analyzed.It also summarizes the causes of train delay,the principle of delay propagation,and the method of delay adjustment.We preprocess and analyze the original data based on the field operation of the railway.Secondly,data are clustered from two perspectives: distribution characteristics and statistical characteristics.The results reflect the law of the changes to the train delay trend,the relationship between delay time and the departure time,the relationship between delay time and the station,and the degree of influence of train delay.Afterwards,train delay status is divided into five categories by clustering,and each of them has obvious characteristics.Thirdly,train delay statuses are classified based on the random forest classification algorithm which has great performance in classification to accurately judge the delay status.Combined with Markov chain,the classification result is taken as the initial state.We construct a Markov transition matrix to analyze the delay trend of subsequent trains.Finally,a train delay adjustment model is established based on the delay status and then solved by using the genetic algorithm.In the delay adjustment model,the total delay and passengers’ dissatisfaction is considered as the objective function,while taking the train dwell time,running time and tracking interval time as constraints.We set the relevant parameters according to the actual operation scenario,then test the model and algorithm using Python programs.The results show that the model proposed in this thesis can effectively reduce the waiting time of passengers while improve the service level of urban rail transit,being compared with the model involving only the total delay time.