Spatial Distribution Pattern And Analysis Of Road Impact Factors For Urban Road Traffic Accident

In recent decades,the rapid development of domestic motorization has witnessed the rapid development of urban traffic accidents.China is the country with the largest number of road traffic accidents in the world,resulting in heavy casualties and huge economic losses.How to reduce the risk of traffic accidents and reduce the severity of traffic accidents has gradually become one of the key issues studied by people.The occurrence of traffic accidents is inextricably linked with people,vehicles,roads,and the environment.The road factor is one of the important aspects,and we can take certain measures to improve it.With the development of traffic informatization,traffic management departments have accumulated a large number of traffic accident data,from the scientific point of view statistics of the occurrence of traffic accidents,analysis of the road influencing factors,so as to propose effective measures to improve the level of urban road safety,The inevitable requirements for ensuring the safety of public transportation.After summarizing the analysis of the spatial distribution patterns and influencing factors of road traffic accidents both at home and abroad,it is found that there are still some problems in the previous studies.First,there are many methods for analyzing spatial patterns of traffic accidents,but some studies still use spatial analysis methods based on Euclidean distance to discuss the distribution patterns of traffic accidents along the road network.In addition,although many studies have conducted statistical analysis on the impact factors of different types of traffic accidents,the existing studies often ignore the influence of traffic autocorrelation on spatial autocorrelation,which makes spatial autocorrelation in the residuals of regressions,thus affecting the regression.result.In view of the problems existing in previous studies,the research in this paper mainly includes the following two aspects:(1)In this paper,based on the urban road traffic accidents,after preprocessing such as coordinate correction,using the road segment as a statistical unit,the network nuclear density estimation method is proposed.Based on the network distance,the distribution density value of eachlixel object is calculated,and a nuclear density estimation and classification map is generated.The results of the analysis show that traffic accidents are clustered on some major roads.In addition,according to the topological relationship between the road segments,the spatial weight matrix is determined,and the calculated global Moran's I statistic is greater than 0,and the P value is less than 0.05.That is to say,traffic accidents have obvious positive spatial correlation in the road network space.Relationships,and when the distance threshold is 300 meters,the Moran's I value takes a maximum value.At this time,the spatial autocorrelation of traffic accidents is themost significant.(2)Because the mixed mode street network in Jianghan District of Wuhan City has both reasonable streets and roads lacking scientific planning,and the urban road infrastructure is relatively backward,the traffic accidents are further aggravated.Therefore,based on the road environment,the paper divides the traffic accident influencing factors into three categories:road structure,road ancillary facilities,and traffic infrastructure.Firstly,it diagnoses multicollinearity problems among various factors,and establishes OLS linear regression models and classic negative binomial terms.The regression model was used for preliminary analysis.Then,based on the spatial autocorrelation of traffic accidents,a spatially lagged variable was introduced to establish a zero-inflated negative binomial regression model.Through comprehensive evaluation and comparison of the models,it is found that the three models of AIC,BIC criteria and Vuong test results all show that the zero-inflated negative binomial regression has a better fitting effect,and its residual Moran's I index also significantly decreases.It shows that the spatial autocorrelation of traffic accidents is eliminated to some extent.The regression results of the zero-inflated and negative binomial model show that the number of lanes,road grades,average speed,road segment connectivity,local depth,road physical isolation,number of bus stops,and distance from the nearest bus station are all significant for urban road traffic accidents.influences.The research results can provide certain scientific references for traffic safety management and smart transportation planning in road environment,and provide technical support for the prevention of traffic accidents.