Study On Spatial Heterogeneity Of Influencing Factors Of Urban Rail Accidents Based On Geographical Weighted Regression Model

Rail transit has become the main choice for urban residents to travel because of its fast,punctual and large capacity.The occurrence of rail transit accidents will seriously affect residents’ travel.Due to the different spatial location of rail lines and stations,there is obvious spatial heterogeneity in the influencing factors of traffic accidents.In order to explore the spatial heterogeneity of the influencing factors of rail transit accidents,this paper analyzes the spatial heterogeneity of the influencing factors of rail transit accidents based on the geographically weighted regression model using the data of Beijing Rail Transit accidents from 2017 to 2018.This paper also digs into the high-risk factors,nodes,and station intervals for the occurrence of urban rail transit accidents,and proposes targeted preventive measures for rail transit accidents.The main research work is as follows:(1)Rail transit accident and influencing factor datasets are constructed.It integrates the rail transit accident data set,weather data set,passenger flow data set of rail stations in and out of stations and passenger flow data set between rail stations from2017 to 2018.This paper also initially identifies 31 variables regarding the impact of rail accidents and retains 21 independent variables after the analysis.(2)The spatial heterogeneity of factors influencing rail accidents is analyzed.Based on the data set of rail transit accidents and influencing factors,using the method of mathematical statistics,this paper analyzes the spatial heterogeneity of the influencing factors of rail transit accidents from the perspective of the temporal and spatial distribution characteristics of accidents,the overall distribution characteristics of accidents and the influencing factors of accidents.(3)Spatial heterogeneity analysis model of influencing factors is constructed.Firstly,through global trend analysis and spatial autocorrelation analysis,the heterogeneity of rail transit accident data set is tested,and it is found that there is spatial heterogeneity.Then,a geographically weighted regression model is constructed to analyze the spatial heterogeneity of the influencing factors of rail transit accidents.And then,based on the original geographical weighted regression model,the geographical variability of influencing factors is tested,and the variables with spatial heterogeneity are selected.Finally,the parameters of influencing factors are calibrated based on semi parametric and multi-scale geographically weighted regression models.(4)The spatial heterogeneity results analysis of the factors influencing rail accidents and suggestions for preventive measures are presented.Compare and select the above model results,analyze the results of the optimal model,and identify the high-risk nodes and high-risk stations.Comparing the evaluation indexes of each model,the goodness of fit of semi parametric geographic weighted regression model is 0.52,AICC is 7128.254,and the fitting effect is the best.For the global variables,identify high-risk factors,and the regression coefficient of "outbound passenger flow five minutes before the accident" is-6.15862,which is a high-risk factor.For local variables,identify high-risk nodes and stations.12 transfer stations such as Haidian Huangzhuang station,Zhichun Road station and national library station are high-risk nodes,and 42 stations such as Fuxingmen Changchun Street and Tiananmen West Tiananmen east are high-risk stations.The rail system shall strengthen safety prevention and control measures between high-risk nodes and high-risk stations to prevent rail transit accidents.