Study On Influencing Factors Of Freeway Accidents

With the acceleration of freeway construction,freeway safety has been widely concerned.It is a necessary way to improve road traffic safety to find the root cause of the accident through in-depth analysis of the internal correlation of data and according to the characteristics of traffic accidents.However,there are slight differences in the data nature of people,vehicles,roads and environment,and different factors have different impacts on different accidents.Therefore,using the same analysis method may lead to inaccurate results.In addition,the correction factor for accident research in China are mostly foreign models and theories,and domestic and foreign in road design,traffic characteristics,such as there are some differences,therefore should be according to the characteristics of road building applicable model for road safety research,at the same time,the traditional model of accident severity research has certain unreliability,to improve the traditional model is necessary.First of all,this paper expounds the research status at home and abroad from three aspects: accident correction factor,heterogeneous modeling and accident severity model,and summarizes the problems that need to be further studied.Then,according to the road design indicators,the road sections are divided into basic units by the homogenous method,the data are analyzed by descriptive statistics,and the abnormal data are processed by the global distance method,and the data are tested by multicollinearity.Secondly,based on the basic laws of accidents and the distribution characteristics of accident data,the basic model applicable to traffic accidents is given.In combination with the homogeneity method,the road sections leading to zero accidents are divided into many characteristics,and the form of zero-expansion negative binomial distribution of accident data is determined by using the Voung statistics for calculation.On the basis of this,the linear correction factor model of the accident is deduced,and the influence of the linear factor on the accident is compared with that of the road in the mountain area and plain area.Then,analyzes the source of heterogeneity in the accident data and to deal with variable form,according to the same variable effects of different accidents randomness of existence and the correlation between variables of random parameter logit model is put forward according to the characteristics of the accident as the dependent variable,people,vehicles,road and environment factors as inde pendent variables to construct the multinomial logit model and random parameter logit model,comparison and analysis the goodness-of-fit of the model and the prediction precision and the heterogeneity of two model parameter estimation results,combined with parameters of the elastic coefficient calculation results for quantitative analysis of the cause.Results show that the logit model considering the heterogeneity of the random variables in the goodness of fit and prediction precision and the heterogeneit y of quantitative ability are better than the fixed parameter model,two models are identified 32 variables has a significant effect on the model,in terms of linear,mountains and plains accident correction factor value larger gap between the distribution of the incident correction factor value direction and position distribution is roughly.The research in this paper is helpful to deeply understand the accident occurrence mechanism and internal law,and provide reference for the construction of accident correction factor model in China.The research results are helpful to make targeted safety improvement strategies to improve road traffic safety.