Modeling And Analysis Of Car- Following Behavior Considering The Asymmetry

In the Car-Following behavior,the phenomenon that the deceleration process is more rapid than the acceleration process is called the asymmetric characteristic.However,most of Car-Following models do not consider this asymmetric characteristic.Few models consider this asymmetric characteristic but can't explain why it exists.The real vehicle trajectory data in the project NGSIM(2004)of the Federal Highway Administration is used as the basic data.According to the data analysis,the asymmetric characteristic in car-following behavior is further confirmed.And the safety factor is the reason why the asymmetric characteristic exists.According to further data analysis,this paper finds that drivers judge the danger of driving process through the change of the space headway and the velocity difference.The mechanism of the asymmetric characteristic formation is given as following.When the space headway is small,drivers pay more attention to the velocity difference,and the deceleration process will be relatively rapid;when the space headway is large,drivers pay less attention to the velocity difference,and the acceleration process will be relatively slow.Based on the Full Velocity Difference(FVD)model,this paper considers that the response of drivers to the velocity difference is negatively correlated with the space headway,and then establishes a new model.In different Car-Following stages,the acceleration analysis indicates that the new model can describe the asymmetric characteristics.The stability analysis further points out that the new model can reflect the asymmetric characteristics.The computer simulation analysis indicates that the new model can describe the asymmetric characteristics on the aspects of density wave,acceleration,hysteresis curve and traffic disturbance,and which is consistent with the real traffic flow.Based on the real vehicle trajectory data,parameters of the new model and the FVD model are calibrated.The result of parameters calibrating shows that the new model can describe the Car-Following behavior more accurately and accurately than the FVD model.