Adjustment Of Train Operation In Urban Rail Transit By Using Global Dual Heuristic Dynamic Programming

In recent years,urban rail transit has developed rapidly.Up to now,the urban rail transit network of many front-line and second tier cities such as Beijing,Guangzhou and Hangzhou has already formed a certain scale.Ensuring normal and timely operation of trains and improving passenger experience are very important indicators in urban rail transit operation.Normally,the subway operators make the planned train diagram.The train runs normally under the planned train diagram,but due to the influence of the passenger flow or other factors,the train operation will deviate from the planned train diagram,especially the rail transit train based on the moving block,which has large traffic density,short tracking interval and delay.The situation is unavoidable,which not only affects passengers’ experience,but also causes large area train delay,which brings trouble to subway operators.The study of adjustment of train operation in urban rail transit is a very effective way to ensure the running of the train,improve the efficiency of the train and reduce the pressure of the operation of the subway operators.In this thesis,on the basis of summarizing the existing train rescheduling models,the train rescheduling model based on LSTM(Long Short-Term Memory)network is put forward by studying the operation control principle of urban rail train and the characteristics of running organization.In this model,LSTM is good at dealing with and predicting the characteristics of relatively long intervals and delays in time series,and the accurate prediction of the delay time of the train arriving at the next station is realized by training the train running parameters by the LSTM prediction unit.Taking the data of Yizhuang Line,Beijing Metro,the LSTM prediction unit in the train rescheduling model has been simulated and verified several times.The results show that the average relative error between the predicted value and the actual value is less than 7%,and the effect is good.According to the train rescheduling model based on LSTM,the characteristics of the model are analyzed and the advantages and disadvantages of the train rescheduling algorithm used in the previous research are summarized,and the method of using the global dual heuristic dynamic programming algorithm to solve the adjustment model is put forward,the algorithm is very applicable to the optimal control of complex dynamic systems such as train operation adjustment,and the balance between control accuracy and computation is balanced.The simulation verification is carried out based on the data of Yizhuang Line,Beijing Metro,and the train delay is divided into two scenarios of initial delay and joint delay.The results show that the adjustment method can achieve the expected goal of eliminating train delay under the two typical train delay scenes.Simulation and comparison between the algorithm and the heuristic dynamic programming algorithm show that the algorithm has obvious advantages in terms of stability and control accuracy.In order to facilitate the analysis of train delay and train operation adjustment effect,"Train Deviation Chromaticity Software" is developed in this thesis.By processing the train operation data,the software realizes the intuitionistic display of the train operation adjustment,and uses the method of operational research to effectively locate the initial delay position of the train.Taking the data of Yizhuang Line,Beijing Metro,the results show that the location accuracy of train initial delay is above 80%.