Research On Severity Prediction Of Highway Traffic Accident Based On Bayesian Network

Nowadays,our country’s economic level has been greatly improved,and the motorization level has also been continuously improved.The number of motor vehicles,highway mileage and motor vehicle drivers have also increased year by year.This has led to the frequent occurrence of highway traffic accidents which has harmed the safety of people’s lives and property.It seriously affected the stable development of society and economy.Therefore,conducting in-depth research on highway traffic accidents,exploring the internal rules between various factors and accidents,and proposing effective safety management methods are very important to reduce accidents and social property damage.This paper takes the classification of accident severity as a starting point,and proposes a highway traffic accident severity prediction model based on Bayesian network to analyze the complex relationship between accident causal variables and accident severity.Firstly,the analysis of highway traffic accidents’ causes was carried out.These causes divided into four aspects:people,vehicles,roads,and environment.Then the severity of accidents was classified into three categories according to the KACBO classification method.The data of some highway accidents in the General Assessments System(GES)are selected as analysis samples.This paper chosen 14 important factor variables and discretized them.Nextly,the structure and parameter learning algorithm are used to obtain the Bayesian network model of highway traffic accident severity prediction with MATLAB software,and the highway traffic accident prediction model based on Bayesian network is constructed by the joint tree algorithm.The model was tested by the model node parameters’ precision accuracy and the model’s prediction hit rate.The results show that the model is effective and its prediction accuracy is highly.Finally,in the instance application phase,the established model was used to predict the severity of highway traffic accidents under variables’ different values,and based on this,the corresponding effective safety measures were put forward.At the same time,by using this model to predict the severity of the highway traffic accidents,the results can provide technical services for relevant administrative departments to formulate scientific and reasonable accident response decisions so as to improve the efficiency of safety management and reduce the damage caused by accidents.