Prediction And Analysis Of Driver Injury In Frontal Crash Based On PSO-BP

In recent years,the car ownership has been on the rise every year,and the following problems such as traffic accidents have become more and more obvious.Therefore,how to reduce the personal injury caused by traffic accidents and rescue the wounded quickly has become a current research hotspot.Based on the real accident cases,this paper deeply analyzes the causes of the accident,explores the influencing factors of the driver’s injury,and uses PSO-BP neural network to establish the driver injury prediction model to quickly predict the driver’s injury after the accident,efficient use of first aid"prime time"to reduce driver mortality.Firstly,the EDR data is introduced,and then the 365 accident cases whose crash type is frontal crash and have EDR data are screened.The information recorded in the accident case is analyzed in detail from the four aspects of people,vehicles,roads and environment,and explore driver injury factors.The results show that when the maximum intrusion value in the cockpit is greater than 30 cm,the proportion of serious injury to the driver increases significantly;The greater the longitudinal maximumΔV,the higher the proportion of serious injuries of drivers;When the driver does not wear safety belt and the airbag is not activated deployed,the proportion of serious injury to the driver is large,accounting for about 62.5%and 66.7%respectively.Then,through the correlation analysis method,it is proved that the longitudinal maximumΔV,seat belt wearing and airbag deployment contained in the EDR data are significantly related to the driver’s injury.Therefore,these three parameters are used as the input of the driver’s injury prediction model.Secondly,Hypermesh and LS-Dyna are used to simulate 100%frontal crash of the vehicle against the rigid wall,and the curve of the acceleration at the driver’s seat with time is obtained.,It is imported into the established Madymo cockpit model as a boundary condition to obtain the driver’s head HIC₃₆ value and chest CTI value under different crash speed and different restraint system parameters.The experiment results are used as the data set required by the driver’s injury prediction model.In addition,the injury results show that the greater the 50ms average acceleration,the higher the driver’s head HIC₃₆ value and chest CTI value.Therefore,the 50ms average acceleration is also used as the input of the driver prediction model,and the output of the model is set as the AIS level corresponding to the driver’s head HIC₃₆ value and the chest CTI value.Finally,the driver injury prediction model is established by using BP neural network and PSO-BP neural network.The results show that the average absolute value error of BP neural network model in predicting head injury value is 11.69%,and the average absolute value error in predicting chest injury value is 12.56%;The average absolute value relative error of PSO-BP neural network model in predicting head injury value is 6.2%,and the average absolute value error in predicting chest injury value is 5.7%.Through comparing the PSO-BP neural network injury prediction model built in this paper,in the 50 accident cases,the prediction results of37 cases are consistent with the real injury situation,the prediction value of 7 cases are higher than the real injury value,the prediction value of 6 cases is are lower than the real injury value,and the AUC value is 0.739,which has a good prediction effect.