Research on influencing factors of major and extra serious traffic accidents based on Bayesian network
Based on the data of 97 major and extra-serious traffic accidents from 2013 to 2021 in China,the relationship between accident impact factors and severity indicators was extracted to explore the key causes of major and extra-serious traffic accidents.Firstly,29 factors containing information about drivers,vehicles,roads,and the environment were extracted from the accident reports.Each category of factors was sorted in descending order of occurrence frequency,and the influencing factors were preliminarily screened based on a cumulative frequency greater than 90％.The Pearson correlation coefficient was used to analyze the correlation between the above factors and the index of accident severity(the number of deaths and injuries in single and multiple-vehicle accidents).Secondly,an improved grey correlation method was proposed to calculate the weighted grey correlation degree of the remaining factors,and the key influencing factors set was constructed by taking the average weighted grey correlation degree greater than 0.75 as the standard.Then,a Bayesian network was established by taking the key factors as node variables.Through the network structure learning based on search scoring and the node conditional probability learning based on Bayesian estimation,14 factors related to accidents were obtained,and 15-factor combination chains were extracted from them.Finally,the risk of single factors was ranked based on interval number theory.The results show that there are differences in the influencing factors of single and multiple-vehicle accidents,with factors such as overloading,large passenger cars,wet roads,no physical isolation,working days,and adverse weather affecting both.However,the degree of impact is different.In addition,fatigued driving and heavy trucks only have an impact on multi-vehicle accidents.Overall,the consequences of multi-vehicle accidents are more severe.The Bayesian network can reflect the true relationship between various factors and has good prediction accuracy.The findings of the study can help the management departments to develop appropriate prevention strategies to reduce the frequency of major and extra serious traffic accidents.
safety engineeringmajor and extra serious traffic accidentsBayesian Network(BN)improved grey correlation analysiskey factor
NETL
NSTL
万方数据
维普
