Research On Decision Support-oriented Car-following Behavior And Dynamic Modeling Method

Transportation system is an important part of modern city,and the decision support based on real-time modeling and simulation of vehicles' operations is becoming the most effective method in the field of traffic management.Traffic simulation system for decision support should be real-time and accurate,while the current one is not.Based on this problem,this pape r makes a deep study of the driving car-following model and proposes a dynamic modeling method that can effectively support decision-making.This paper proposes a method of driving modeling using the driver's mobile phone to collect real-time location information,and demonstrate the feasibility of it through the car-following test on actual road.The simple algorithm is used to calibrate the extended GM model,and to comprehensively evaluate the fitness,robustness,and the sensitivity of parameters of the model,therefore the feasibility of the data with noise in the modeling of car-following behavior was verified.Driving characteristics of different drivers are compared car-following tests with simulation software.The drivers' reaction time in different stages is calibrated,which proves the data obey a certain multimodal logarithmic normal distribution.Based on the traffic data,the parameters is calibrated with genetic algorithm.The distribution of the model parameters corresponding to different people is analyzed with the method of the cluster analysis.The method is preliminary verified through the evacuation simulation system,and it is confirmed that whether to consider the individual heterogeneity in the modeling process will have effect on the final macro simulation results.This paper demonstrates the feasibility of modeling the driving behavior of mobile data and the heterogeneity of driver's driving behavior by experiments in actual and simulation,and proposes an initial framework of traffic simulation model for decision support.The research results of this paper can provide scientific and theoretical support for the development of traffic simulation system.