Urban Rail Transit Passenger Flow Change Trend Prediction And Traffic Dispatching Optimization

With the continuous development and improvement of urban rail transit system,people gradually take urban rail transit as the preferred mode when they travel,and the requirements and standards of urban rail transit operation and management are continuously improved.Considering that in the actual operation management process,the passenger flow has the characteristic of strong random fluctuation,the division of the operation time is difficult to reflect the actual passenger flow change situation,so the transaction record of passenger inbound and outbound produced by the automatic ticketing system(AFC)in the urban rail transit system are used to seek a method which can characterize the change trend of line passenger flow,and studies the optimization of traffic dispatching in urban rail transit as a reference basis.Firstly,the research status of short-time passenger flow prediction and bus dispatching optimization is introduced.Then,the automatic ticketing system,the development status of Nanjing Metro and the data collection of AFC passenger flow are introduced,and two ideas of the trend of passenger flow are put forward: one is to predict the passenger flow first,and then to analyze the changing trend of the passenger flow in the next moment.The corresponding training data and test data are prepared according to the requirements of direct prediction and indirect prediction model of passenger flow change trend.Then,based on the passenger flow data set of all stations of the preprocessed target line,the realization scheme of the forecast of passenger flow change trend are analyzed,and the site with typical characteristics of two passenger flow in Zhujiang Road station and Nanjing South railway station are selectd as an example,based on the pre-extracted traffic data set of the inbound and outbound station of the first line,the index and calculation method of passenger flow trend characterization are analyzed,and then Python and Matlab are used to calculate the traffic statistics description results and passenger flow trend of all stations in the designated Operation Day(2017-10-01~2017-10-30)line.Finally,taking the typical station of October 27,2017-Zhujiang Road station and Nanjing South Railway station as examples,the realization of the research scheme of passenger flow change trend are introduced.At the same time,the corresponding change trend prediction results are obtained by using the same scheme steps for all sites of the target line,and the data mining technology is used to cluster the prediction results of passenger flow trends,so as to obtain the result of operation time division which is more in line with the daily traffic change of rail transit.By using Fisher optimal segmentation method,the state sequences of the prediction results of two kinds of passenger flow trends are segmented,and the loss of the prediction results obtained by the two methods is calculated to compare the performance of the two ideas.Comparing the optimal segmentation results of two kinds of ideas,it can be concluded that the loss function value of the forecast result of passenger flow change trend based on Markov model is small,and the distribution of boundary points is more uniform.Choosing a better Division scheme combined with the preprocessed line passenger flow data set,the driving scheduling optimization model is established with the starting time as the decision variable,and the genetic algorithm is used to solve the model,and the optimized departure schedule is obtained,and the average time length of passenger waiting is taken as the evaluation index,Compare it with the actual operation scheduling schedule the overall waiting time of a single train carrying passengers,for most shift trains,the length of the scheduling optimization model is lower than the actual scheduling scheme,which verifies the effectiveness of the optimization scheme.In a word,based on the historical passenger flow data produced by AFC system,a method to analyze the change of passenger flow and predict its trend is discussed which provides a different way to improve the quality and efficiency of rail transit operation management and enhance the passengers' experience.