Research On Dynamic Platoon Dispersion Model In A Connected Vehicle Environment

The vehicle arrival distribution at the intersection is an important input of the adaptive signal control system.The parameters of the traditional static platoon dispersion model are based on historical data and cannot reflect the dynamic characteristics of the traffic flow well.The development of IOV(Internet of Vehicle)technology has brought new opportunities to solve the problem.Under the vehicle-connected environment,data,such as vehicle serial number,travel time and travel speed,are easily obtained.Based on this,a dynamic platoon dispersion model reflecting the real-time changes of traffic flow can be constructed,which can be the reflection of the dynamic variation characteristics of traffic,and the results can be used for adaptive signal timing optimization.Hence,on the premise of vehicle-connected environment,this paper studies the relative characteristics of dynamic platoon dispersion model and compares it with static platoon dispersion model to verify its prediction effect.The main work are as follows:(1)The relative characteristics of platoon dispersion model are studied and the classic platoon dispersion models,Pacey model and Robertson model,are introduced.Furthermore,the influencing factors of platoon dispersion law are analyzed.(2)This paper introduces the acquisition method and key points of the data required by platoon dispersion model.A case study of the platoon data of Jiulidi East Road in Chengdu,this paper analyzes the relationship between the discrete characteristics of platoon and the link length and traffic load.Meanwhile,it verifies the urban traffic flow and demonstrates the heterogeneity of urban traffic flow;according to various distribution fitting for vehicle travel time data,the paper validates the rationality that the travel time in platoon dispersion model obeys normal distribution,truncated normal distribution,lognormal distribution and truncated lognormal distribution,which lays the foundation for follow-up research on dynamic platoon dispersion.(3)Static and dynamic platoon dispersion model is built respectively based on travel time that obeys four probability distributions,that is,normal distribution,truncated normal distribution,lognormal distribution and truncated lognormal distribution.Taking vehicle travel time that obeys truncated normal distribution as an example,this paper takes theoretical derivation both for static and dynamic platoon dispersion model.At the same time,a dynamic Robertson platoon dispersion model is constructed and based on MATLAB,a numerical calculation method of platoon dispersion model is established.(4)Ten scenarios are made up according to the survey data of five platoon dispersion models in non-peak hours and peak hours respectively.And it compares the prediction effect between static and dynamic platoon dispersion model.The results show that comparing to static platoon dispersion model,the average prediction error of the five dynamic models is reduced by 21.78% in non-peak hours and 3.16% in peak hours.Moreover,in different settings,the prediction results of dynamic platoon dispersion model are superior to the ones of static platoon dispersion model,which indicates that the prediction results of dynamic platoon dispersion model are more accurate and dynamic model can also adapt to real-time traffic flow changes better.