Research On Data Driven Vehicle Microscopic Simulation Model

Traffic simulation system is an efficient method to study the traffic problem.Using macroscopic,macroscopic or microscopic traffic model reproduce the traffic scene.Traffic simulation is a basic element of traffic simulation,in the microscopic simulation model,the behavior of the vehicle is the main body of the description,and the driving behavior of the vehicle on the road is described by the car-following model and lane changing model.In the past decades,many improved car-following models based on the full velocity difference(FVD)model have been developed.But these models do not consider the acceleration of leading vehicle.Some car-following models considered individual anticipation behavior of drivers,However,their classification of driving types is subjective and not based on actual traffic data.As an important model of traffic model,lane changing model is one of the most important research subjects in the field of transportation.In recent years,there are a large number of discretionary lane changing model based on machine learning have been proposed,but these discretionary models are classifying all the lane changing data directly,without considering the influence of different driving styles on the lane changing model.But in real life,the driver due to age,driving proficiency,physical quality and other reasons will show different driving styles.The prediction accuracy of existing models for lane changing events is low.In this background,this paper presents a new car following model and lane changing model based on the machine learning algorithm,and develops a new microscopic traffic simulation system.The main contents and contributions of this paper are as follows:(1)A new car-following considering the driving style and the acceleration information of leading vehicle is proposed.Driver’s driving style is categorized based on actual traffic data via the method of clustering algorithm,the proposed car-following model on the basis of FVD model is developed taking into account individual anticipation effects and the acceleration of leading vehicle.The stability condition of traffic flow is obtained by linear stability analysis and the effect of considering driving characteristics and leading vehicle’s acceleration on car-following behavior is analyzed via numerical simulation.(2)A discretionary lane changing model considering driving style is proposed.The influence of driving type on the discretionary lane changing model is considered based on the K-means clustering algorithm.And the multilayer perceptron classifier is used to predict the discretionary lane changing events.Finally,according to the results of the comparison between the prediction accuracy of the new model and other models,it can be seen that the discretionary lane changing model considering the driving style can greatly improve the prediction accuracy of lane changing events.(3)Based on the proposed car-following model and lane changing model,a 3D microscopic traffic simulation system based on data driven is developed,the traffic flow of Beijing Capital Airport Expressway and Beijing South Railway Station is simulated,and the interactive control function is provided.In this paper,a 3D microscopic traffic simulation system based on data driven is realized,according to experimental results,the car-following model proposed in this paper improves the stability of traffic flow,the new lane changing model has a higher prediction accuracy.