Study On Influencing Factors Of Pedestrian Injury Based On Accident Data

The road traffic safety situation in China is very severe and the number of traffic accidents is high every year,tens of thousands of people being injured or killed in traffic accidents;Pedestrians,as weak participants in road traffic,have always occupied a high proportion of casualties in traffic accidents.Compared with occupants in motor vehicles,pedestrians are more vulnerable after collision with vehicles.Pedestrian collision safety has always been a research hotspot in the field of road traffic safety in China.Based on the background,the paper takes the influencing factors of pedestrian injury in human-vehicle collision accidents in China as the research content,and obtains the real traffic accident data which can reflect the human-vehicle collision accidents in China through sampling investigation and in-depth investigation of road traffic accidents in China.Through screening the collected accident data,find out the typical cases of human-vehicle collision accidents,and make statistical analysis on the case data,getting the distribution characteristics of influencing factors of pedestrian injury in human-vehicle collision accidents in China;Logistic regression analysis and cluster analysis are used to find out the correlation between pedestrian injury and various influencing factors in traffic accidents in China.The prediction model of pedestrian injury in accident is constructed by using BP neural network model,and the severity of pedestrian injury in accident is inferred according to the distribution of influencing factors in accident scene.According to the statistical analysis of influencing factors of pedestrian injury and the correlation analysis,the prediction analysis of pedestrian injury,countermeasures and suggestions for pedestrian protection technology are put forward.The research contents include:Studying and analyzing the sampling method,the cross variable quota sampling is selected to extract and collect samples from the national traffic accident data;Taking the collision between people and vehicles as an example,this paper introduces the methods and processes used in the in-depth investigation of accidents,including the data collection of accident scene,the information collection of accident vehicles and personnel,the analysis of personnel damage,etc.By comparing the collected accident data with the annual report data of road traffic statistics in China,it is found that the collected accident data can fit well with the road traffic accident data in China,which proves the reliability and representativeness of the collected accident data.359 typical cases of human-vehicle collision accidents in China are studied and analyzed.The pedestrian injury severity,pedestrian information,vehicle information,road information and accident characteristic information(16 related variables)are statistically analyzed,obtaining the distribution characteristics of influencing factors of pedestrian injury in humanvehicle collision accidents in China.So the basic characteristics of human-vehicle collision accidents in China can be obtained.Logistic regression analysis and cluster analysis are used to study the correlation between pedestrian injury and influencing factors.Logistic regression analysis can find out the significant influence degree of each influencing factor on pedestrian injury through goodness of fit.Cluster analysis studies the correlation between pedestrian injury and each factor level by clustering pedestrian injury and influencing factors,finding out the factors and levels related to pedestrian injury.The prediction model of pedestrian injury degree is established by using BP neural network.Ten factors such as pedestrian and vehicle are taken as input variables,and pedestrian injury degree is taken as output variable.The model is trained by the selected data of human-vehicle collision accident.Finally,the model is predicted and verified,and it is found that the model can accurately and effectively predict pedestrian injury degree according to the factor information in the accident,predicting the injury degree of travelers by accident scene factors.According to the distribution characteristics of influencing factors of pedestrian injury,the correlation analysis and prediction results between pedestrian injury and each influencing factor,the influence of different factors and related levels on pedestrian injury degree is obtained.According to the research conclusion,the improvement countermeasures of pedestrian protection technology are put forward from three aspects of vehicle,pedestrian and traffic management to reduce pedestrian injury degree in accidents.