Optimization Of Urban Rail Transit Timetable Based On AFC Data Mining

Urban rail transit has become the core of public transport in maor cities with the expansion of population size and the gradual formation of rail transit network.Timetable,which is an important part of the operation and management of rail transit,has a direct impact on the transport capacity and the operating costs.A large amount of passenger travel OD data is stored in the database of the AFC system,which is the natural material for studying the characteristics of passenger flow behavior and the preparation of the timetable.This paper designs a method to extract the passenger flow OD data from the AFC database,and makes an in-depth study on the timetable optimization model and the solution method.This paper firstly introduces the research status at home and abroad about the preparation of the timetable,and expounds the main content and technical route of this paper.Then,the structure of AFC system and the travel OD data format stored in the database are introduced,and the travel OD data of Beijing Subway is pretreated.Based on the post-treatment OD data,this paper researches the time and space characteristics of the passenger flow,and analyzes it with the Beijing Subway.Then,the "entire network OD" and "associated OD" concept are put forward,and the shortest path search algorithm and passenger flow distribution are introduced.The space-time correction model is put forward,and the travel OD data in the entire network is distributed to the target line of Beijing Subway.The OD data associated with the target line is extracted.After that,the associated data is calculated,the arrival rate matrix and departure rate matrix are obtained,and the number of people who arrive and leave the station is obtained by using the three spline interpolation.Design the passenger flow characteristic matrix during the day and in the day,and based on the passenger flow characteristic matrix during the day,use hierarchical clustering method and the k-means clustering method to cluster the operation day and divide it into working days and no working days.At the same time,the optimal cutting method is used for the operation time clustering based on the intraday characteristic matrix,and the working days are divided into five operational periods.Finally,based on the division operation periods,set departure interval in different operation periods as decision variables,a scheduling optimization model is proposed which involves the passenger waiting time,the crowded degree in train and the operating costs of the operation company.Based on the training set.the genetic algorithm is used to train the model to solve the problem,and the optimized interval in different operation periods will be obtained.Then we test the results on the test set,the optimized timetable scheduling indicators are better than carriers schedules which is used now.Therefore,the validity of the optimization results will be verified,and new idea will be provided for the operation company to compile timetable.