Research On Influencing Factors Of Traffic Accident Severity Of Elderly Drivers

Due to the increasing number of road traffic accidents caused by motorization and the increasing number of elderly drivers under the background of aging population,it is necessary to study the traffic accident severity for elderly drivers.In this paper,the data set released by the National Safety Administration of the United States was fused and cleaned to conduct a comparative analysis of the factors influencing the accident severity in different age groups and an in-depth analysis of elderly drivers,so as to provide a basis for road safety management of elderly drivers.The main work of the study is as follows:(1)Analysis of influencing factors of accident severity based on contingency table analysis.It is found that the accident severity of drivers in different age groups is significantly different by using contingency table analysis.Further analysis shows that the accident severity of elderly drivers is correlated with driver gender and collision type,but not with weather type.The influencing factors such as gender and collision type should be focused on and further studied.(2)Comparative analysis of influencing factors of drivers in different ages based on Random Parameter Logit model with mean heterogeneity.The results show that the random parameter variable of rear-end collision has mean heterogeneity in the elderly variable,and elderly drivers will reduce the severity of rear-end collision.The horizontal direction,vertical slope and intersection type of road have significant influence on the severity of accidents of elderly drivers,but not for young and middle-aged drivers.When the road is curved,the probability of serious injury will increase by 0.4%,the probability of minor injury will increase by 0.4%.When the road is sloping,the probability of serious injury will increase by 0.4%.When the road is non-intersection,the probability of serious injury will increase by 1.4%.The number of passengers increased the probability of serious injury by 0.9% for young drivers,but not for older drivers.The probability of serious injury of middle-aged drivers decreased by 0.4% during the peak period,but not for older drivers.(3)Prediction of accident severity and importance ranking of influencing factors of elderly drivers based on deep learning.Combining the random forest algorithm with convolutional neural network,the convolutional neural network based on characteristic gray scale image can improve the prediction accuracy and the model is advanced.The model results show that the significant factors influencing the accident severity of elderly drivers are as follows: collision type,gender,vehicle type and road speed limit.Head-on collision,female driver,car and road speed limit of 30-60km/h all increase the accident severity of elderly drivers.Therefore,safety training and education on driving risks such as head-on collision and speed limit of 30-60km/h for elderly female car drivers should be strengthened.