Predictive analysis on anxiety level of elderly drivers in different conflict situations
To investigate the anxiety levels of elderly drivers in the conflict scenarios,a scale was employed to quantify the anxiety degree of elderly drivers.A virtual scenario of conflict intersection road was constructed to collect the driving behavior data of elderly drivers in different various conflict situations.The Spearman correlation analysis method was utilized to screen out the factors influencing the anxiety levels of elderly drivers.The prediction models for the anxiety levels of elderly drivers were established by using radial basis function(RBF)neural network and backpropagation(BP)neural network respective-ly,and the prediction performance of the two models was compared.The results show that in different conflict situations,the age,driving years,brake pedal depth,steering wheel angle entropy,and conflict severity present the significant positive corre-lation with the anxiety level,while the speed presents a significant negative correlation with the anxiety level.The prediction accuracy of the elderly driver anxiety model based on RBF neural network is 87.14％,the accuracy rate is 88.24％,the re-call rate is 68.18％,and the F1 value is 76.92％.The prediction accuracy of the elderly driver anxiety model based on BP neural network is 92.86％,the accuracy rate is 90.48％,the recall rate is 83.36％,and the F,value is 88.37％.Both mod-els can better predict the anxiety level of elderly drivers,and the anxiety prediction model of elderly drivers based on BP neu-ral network has better prediction performance.The research results can provide a theoretical basis for correctly identifying the anxiety level of elderly drivers,and are of great significance for creating the safe and efficient driving.
万方数据
维普
