基于前后多车信息的跟驰模型及其车流平稳性控制研究

发布时间：2018-05-30 12:09

本文选题：交通流 + 跟驰模型　；参考：《重庆大学》2014年博士论文

【摘要】：随着道路交通流量的不断增加，车辆之间的作用关系越来越明显，刻画车辆作用关系的内在机理，揭示其演变规律是提高车流平稳性的重要手段。ITS(IntelligentTransport System)系统的广泛深入应用，使得车辆可快速获取与反馈运行信息，并根据这些信息合理控制自身状态，以达到前后多车的协同行驶，从而形成一个有序流动的车辆行驶队列。然而，传统的交通流模型未考虑前后多车信息，已难以刻画车辆相互作用下的车流演变规律。因此，系统地刻画ITS环境下车流演变规律，进而研究提高车流稳定性的控制方法已成为亟待攻克的核心关键任务。为此，本文以现有交通流微观模型为基础，考虑前后多车信息，建立车辆相互作用下的跟驰模型，以揭示ITS环境下车流演变规律；在此基础上，为使模型能适应交通系统非线性变化的特点，对模型进行时变参数标定；进而研究保证车流平稳运行的控制策略。论文的主要工作如下： ①考虑邻近双前车最优速度差及其后视效应的综合作用，构建基于前后车信息的CI-CF跟驰模型，揭示考虑前后多车信息的车流演变规律。基于FVD(full velocity difference)模型、BLVD(backward looking and velocitydifference)模型，，利用在ITS环境下获得的前后车信息，提出一个考虑邻近双前车最优速度差和后视效应CI-CF(a new comprehensive information car-following)模型。通过线性稳定性分析，得到模型的稳定性判据；采用非线性分析方法，推导出mKdV方程，用于描述系统在临界稳定点附近的交通流特性。在周期边界条件下，运用数值仿真验证车辆启动过程、停止过程、演化过程的理论研究结果正确性。仿真结果表明，与FVD模型、BLVD模型相比，综合考虑邻近双前车的最优速度差和后视效应，可使车流拥有更好的协同行驶特性，最大程度使得车流运行行为趋于一致，真实刻画了车流演变规律。 ②针对跟驰模型时变参数标定问题，提出一种自校正的参数标定方法，使CI-CF模型更能适应交通时变特性。为使CI-CF模型能更加准确地刻画各类交通非线性现象，必须应用实测交通数据对模型进行参数标定。首先给出基于最小二乘法的模型参数标定方法；考虑到交通系统时变特性，然后提出一种自校正的跟驰模型参数标定方法，以解决跟驰模型时变参数的标定问题。实验结果表明，提出的自校正参数标定方法因可以根据交通环境的变化来实时动态标定参数，与传统成批处理的最小二乘参数标定法得出一个恒定的参数标定结果相比，更能准确刻画车流演变规律。利用提出的参数标定方法对FVD模型、BLVD模型进行参数标定，实验结果进一步证明CI-CF模型能更准确刻画车流演变规律。 ③针对车流平稳性控制问题，考虑多前车稳态期望速度效应，基于CI-CF模型，提出保证车流平稳运行的控制策略，形成MSDVE模型。在Konishi等人的基础上，充分考虑前方多辆车的信息，以驾驶员期望的稳定速度前进为目标，基于CI-CF模型，设计一个车流平稳运行的控制策略，形成MSDVE(multiple steady desired velocity effect)模型。运用反馈控制理论，得到车流保持稳定的条件。仿真结果表明，在同等条件下，提出的模型所得到的车流平稳性运行状况优于无控制情况，也优于KKH(Konishi K.,et al)模型、ZG(Zhao and Gao)模型的结果。与ITS(Han X L)模型相比，由于考虑了稳定期望速度效应，可使交通系统的稳定性增强，车流运行更加平稳。 ④在上述平稳性控制策略基础上，进一步考虑后视效应的作用，提出基于前后车综合信息的稳态期望速度控制策略，形成SDVEPF模型。在考虑多前车稳态期望速度效应的稳定性控制策略基础上，进一步考虑后视效应的影响，提出考虑前后车综合信息的车流平稳性控制策略，形成SDVEPF(steady desired Velocity effect of Preceding and Following Cars)模型。同样运用反馈控制理论，得到车流保持稳定的条件。仿真结果表明，考虑前后车稳态期望速度综合效应作用，车流运行状态更加平稳，车辆速度平滑度和幅度更小，因此，考虑前后车稳态期望速度综合效应的SDVEPF模型，比仅考虑多前车稳态期望速度效应的MSDVE模型更利于控制车流平稳运行。综上所述，本文立足于交通系统智能化，提出了更符合交通实际和具有一定前瞻性的跟驰模型，研究了模型线性与非线性特性、时变参数标定以及车流平稳性控制问题，理论分析和仿真实验均验证了上述工作的有效性。研究成果为ITS环境中车流运行状况的刻画与分析，提供了相关理论基础和方法。
[Abstract]:With the increasing traffic flow , the relationship between vehicle and vehicle becomes more and more obvious , the internal mechanism of vehicle interaction is more and more obvious , and its evolution law is revealed as an important means to improve the stability of traffic flow .

In this paper , based on the existing microscopic model of traffic flow , the following vehicle information is taken into consideration , and the following model is established under the interaction of vehicle to reveal the evolution law of traffic flow in ITS environment ;
On this basis , in order to adapt the model to the nonlinear change of the traffic system , the time - varying parameter calibration is carried out on the model ;
Furthermore , the control strategy to ensure the smooth running of traffic flow is studied . The main work of this paper is as follows :

In this paper , the CI - CF following model based on the front and rear vehicle information is constructed in consideration of the comprehensive effects of the optimal speed difference and the rear - view effect of the two - front vehicle .

Based on the FVD ( full velocity difference ) model , BLVD ( backward looking and velocitydifference ) model , a new comprehensive information car - following model is proposed considering the optimal speed difference and the rear - view effect CI - CF ( a new comprehensive information car - following ) model in ITS environment .
The nonlinear analysis method is used to derive the mDk equation , which is used to describe the traffic flow characteristics of the system in the vicinity of critical stable point . The simulation results show that the optimal speed difference and the rear - view effect of the vehicle start - up process , the stopping process and the evolution process are verified by using the numerical simulation under the periodic boundary condition . The simulation results show that the optimal speed difference and the rear - view effect of the adjacent double - front vehicle are comprehensively considered in comparison with the FVD model and the BLVD model , so that the running behavior of the traffic flow tends to be consistent , and the evolvement rule of the traffic flow is vividly portrayed .

( 2 ) Aiming at the problem of time - varying parameter calibration of the following model , a self - correcting parameter calibration method is proposed , so that the CI - CF model can adapt to the time - varying characteristics of traffic .

In order to make CI - CF model more accurately depict various traffic nonlinear phenomena , it is necessary to use the measured traffic data to calibrate the model . First , the calibration method of model parameters based on least square method is given .
Considering the time - varying characteristics of traffic system , a self - correcting parameter calibration method is proposed to solve the problem of calibration of the time - varying parameters of the following model . The experimental results show that the proposed self - tuning parameter calibration method can accurately depict the evolvement rule of traffic flow according to the change of traffic environment . The parameter calibration method is used to calibrate the FVD model and BLVD model . The experimental results show that the CI - CF model can more accurately depict the evolution law of traffic flow .

( 3 ) Based on CI - CF model , based on CI - CF model , this paper puts forward a control strategy to ensure the smooth running of traffic flow based on CI - CF model . The MSDVE model is formed .

On the basis of Konishi and so on , taking fully into consideration the information of the front multi - car , taking the steady speed expected by the driver as the target , based on the CI - CF model , we design a control strategy for the steady operation of the vehicle flow . The simulation results show that under the same condition , the traffic stability is better than that of the model of KKH ( Konishi K . , et al ) and ZG ( Zhao and Gao ) . The simulation results show that the stability of the traffic system can be enhanced and the traffic flow is more stable than the ITS ( Han X L ) model .

On the basis of the above - mentioned stationarity control strategy , the effect of the back - view effect is further considered , and a steady - state expected speed control strategy based on the comprehensive information of the front and rear vehicles is proposed , and an SDVEPF model is formed .

On the basis of the stability control strategy considering the steady - state expected speed effect of the multi - front vehicle , the influence of the rear - view effect is further considered , and the steady desired velocity effect of the vehicle flow is obtained . The simulation results show that the SDVEPF model considering the comprehensive effect of the steady - state expected speed of the front - rear vehicle is more stable and the vehicle speed is smooth and the amplitude is smaller . Therefore , the model of the SDVEPF considering the comprehensive effect of the steady - state expected speed of the front and rear vehicles is considered , which is more beneficial to control the smooth operation of the traffic flow than the MSDVE model considering the steady - state expected speed effect of the front and rear vehicles .

In conclusion , based on the intelligence of traffic system , this paper puts forward a model which is more consistent with traffic reality and a certain forward - looking model . The linear and nonlinear characteristics of the model , the calibration of time - varying parameters and the stability control of traffic flow are studied , and the theoretical analysis and simulation experiments verify the validity of the above work . The research results provide the theoretical basis and method for the characterization and analysis of traffic flow health in ITS environment .
【学位授予单位】：重庆大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：U491.1;U495;TP13

【参考文献】