Research On Urban Rail Transit Passenger Flow Characteristics And Weighted Network Invulnerability Based On AFC Data

Due to its large capacity,fast speed,and environmental protection,urban rail transit not only increases annual passenger traffic year by year,but also accounts for an increasing proportion in the urban passenger transport system.Rail transit has gradually become the preferred mode of transportation for residents.However,the rail transit network that gathers a large number of passenger flows is fragile.The rail transit system will cause station failure due to weather or equipment failures.The rail network structure and passenger flow will be greatly affected.Line network planning and dynamic operation of rail transit are of great significance.This paper takes the short-term prediction results of OD passenger flow as the weight of the passenger flow of the track network to study the network characteristics of the track network in different time periods.The main research contents are as follows:(1)The data processing method of subway card swipe is described,based on the data of subway card swipe The overall passenger flow characteristics,line passenger flow characteristics,OD passenger flow time characteristics,and station classification according to the station passenger flow characteristics are analyzed;(2)The passenger flow prediction theory of the passenger flow operation stage is elaborated,and the machine learning theory is introduced into the short-term passenger flow prediction theory,and the Station OD passenger flow similarity calculation method;(3)Based on the subway normal passenger flow characteristics and the best time granularity,construct a short-term passenger flow prediction model based on the XGBoost model to achieve OD passenger flow prediction;(4)based on the results of the short-term passenger flow prediction model,build The weighted complex network of passenger flow is studied,and the network characteristics of the three periods of morning peak,flat peak and evening peak are studied and important stations are identified;(5)According to the classification of subway operation accidents,the network faults of different types of stations in the morning and evening peak are calculated respectively Invulnerability.The results show that(1)Hangzhou rail transit stations can be divided into 6 categories according to passenger flow characteristics;(2)The optimal time granularity of Hangzhou OD passenger flow prediction is 10 minutes,and the factor that has the greatest impact on the prediction is the standard operation of trains between stations Time,site type and lagging characteristics;(3)The short-term passenger flow prediction model based on XGBoost has high accuracy.The MAE of the entire network OD passenger flow prediction is 10.5384 and the RMSE is 21.159,which is better than the traditional cubic exponential smoothing method.Great improvement;(4)The weighted network characteristics of passenger flow in different time periods have certain differences,and the sequence arrangement of important stations is quite different;(5)Station failures do not affect train traffic failures,and the loss rate of passenger flow at early peaks It is greater than the evening peak,and the impact on adjacent stations is related to the distribution of line passenger flow;station failure affects train traffic failures compared to the network.The Hangzhou rail transit network is less resistant to such failures than the previous one Class failure.