Study On The Car Following Model And Its Smoothness Control Based On The Information Of Preceding And Follwing Vehicles

With the increasing traffic flow, the relationship between vehicles is more and moreobvious. It is very important to describe the internal mechanism of the relationshipbetween vehicles. Moreover, revealing the evolvement rule about traffic flow is animportant means to improve the stability of traffic flow. Meanwhile，with developmentof ITS, the dynamic and real-time information interactions among vehicles becomefeasible. With the aid of such interaction, drivers can control their speeds to realizecooperative driving with the following and preceding vehicles beyond their vision, andthus to form a queue of vehicles moving ahead in an orderly manner. However，traditional traffic flow models usually don’t take the following and preceding vehiclesinto consideration，and they are unsuitable to describe the traffic rules in cooperativedriving traffic flow and complex relation. Therefore, describing traffic rules andimproving traffic flow stability are the key problem needs to be solved first.Based on the existing research of traffic flow, under the interaction of traffic flow，a microscopic traffic model is set up by considering the comprehensive information ofthe following and preceding vehicles. And on the basis, the model’s parametercalibration is studied by using the measured data. Then, the work of parametercalibration can make the model be more aligned with actual traffic environmentcharacteristics of space and time. At last, how to use the model in traffic congestioncontrol is investigated. The main work is as follows:①Based on the combined effect of the optimal velocity difference information oftwo neighboring preceding vehicles and their effect of looking backward, a newcomprehensive information car-following model is established, which is designed toreveal the evolvement rule about traffic flow.Based on FVD (full velocity difference) model and BLVD (backward looking andvelocity difference) model, a new comprehensive information car-following (CI-CF)model is proposed by taking advantage of the dynamic and real-time informationinteractions among vehicles under the ITS environment. So, the CI-CF model takes fullaccount of the optimal velocity difference information of two adjacent precedingvehicles and their effect of looking backward. The stability criterion of the model isgiven through the linear stability analysis. The mKdV equation which describes itspropagation pattern of density wave near the critical point is derived through nonlinear analysis. Under the periodic boundary conditions, the accuracy of the theory is studiedby using the numerical simulation to verify the vehicle start-up and stop process.Numerical simulation shows that, compared with FVD model and BLVD model, ourmodel is reasonable which simultaneously takes the optimal speed difference of twoadjacent preceding vehicles and looking-backward effect into account. It can be used toachieve the best synergy drive traffic, effectively reduce adverse effects within thesystem, make the traffic operation behavior consistent at the greatest extent, improvethe stability of traffic flow, and objectively reflect the traffic flow.②For time-varying parameter calibration of CI-CF model, a self-tuning methodof parameter calibration is proposed, which is designed to make CI-CF model betteradapt to the time-varying traffic characteristics.In order to make all kinds of traffic nonlinear phenomena described accurately byCI-CF model, model parameter calibration must be done by using actual data cases atfirst. Based on least square method, a method of model parameter calibration is given.Meanwhile, by taking transport system’s time-varying characteristics into consideration,a self-tuning method of parameter calibration is proposed, which is designed to solvethe problem of time-varying parameter calibration of car following model. Theexperiment results show that compared with constant parameters calibration results byusing the traditional batch processing least squares parameters calibration method, theCI-CF model preferably reflects the real traffic flow evolution law because of modelparameters changing with the change of the traffic scene real-time dynamic. And thenthe FVD and BLVD model parameters are estimated by means of the proposed method.The experiment results further prove the perfection of CI-CF model.③For traffic flow stability control, the MSDVE model, the feedback control oftraffic jam based on car-following model is discussed based on the steady stateexpectation speed effect, which is designed to ensure the traffic running smoothly andsteadily.Based on the Konishi’s work, the MSDVE (multiple steady desired velocity effect)is proposed based on CI-CF, which fully takes account of the information of many carsin front of experimental vehicle and aims for drivers’ expect stable speed. The feedbackcontrol of traffic jam based on car-following model is investigated from the system andcontrol viewpoint. Numerical simulation shows that the results of the model proposed inthis paper is better than those of without control and KKH (Konishi K., Kokame H. andHirata K.) model and ZG (Zhao and Gao) model. Compared with ITS (Intelligent Transport System, Han X L), our model can make the system increases, the stability ofthe vehicle speed more smoothly because of taking account of Constant rate expectedeffect. As a consequence, the feedback control of traffic jam effectively alleviateexpressway congestion situation, and increase the utilization of expressway.④Based on above-mentioned control policy, the influence of back lattice(backward-looking effect) on traffic flow is incorporated, and then the SDVEPF modelis presented which is based on control policy established on the comprehensiveinformation of the front vehicle and the following vehicle in the traffic low.Based on preliminary work, the influence of back lattice (backward-looking effect)on traffic flow is incorporated and then a SDVEPF (steady desired Velocity effect ofPreceding and Following Cars) model is presented. Traffic flow stability conditions aregiven on the basis of the theory of feedback control. Simulation results show that, byconsidering the backward-looking effect, both the propagation speed and amplitude ofdensity wave are decreased, and thus the stability of traffic flow is enhanced. As aconsequence, based on steady expectation speed before and after comprehensive effect,the method has a better congestion inhibition performance than other congestion controlmethod.In conclusion, the car-following model is proposed based on the intelligenttransportation system, and the model is far-sighed and has more realistic congestion andflow patterns. In addition, model’s linear and nonlinear properties, the work ofparameter calibration and the feedback control of traffic jam are studied, and in the end,theoretical analysis and simulation results validate the effectiveness of the work.