考虑交通环境效应的动态次优拥挤收费研究

The proposed project aims to develop comprehensive models to support decision making of various forms of dynamic second-best congestion pricing considering environmental externality.. For the three main forms of second-best congestion pricing, namely the cordon pricing, area pricing and HOT lane pricing, the corresponding dynamic second-best congestion pricing problems are first formulated as mathematical programs with equilibrium constraint, where the requirements on different toll strategies including the toll locations are specified as different constraint sets, and travelers are assumed to follow dynamic user equilibrium, which is written in form of variational inequality in the problems.. One of the mainstream algorithms for solving dynamic traffic assignment depends on the cell-transmission based flow propagation. Utilizing the traffic information carried in each cell, a modal sensitive vehicle emission model is developed to compute the time dependent emission rate for each cell. Subsequently, another cell-based non-steady-state air dispersion model is built to capture the dispersion of air pollutants and compute the roadside pollutant concentrations, which actually decide the local air quality. Considering the uncertainty of the meteorological conditions during the implementation of traffic management measures, a risk-averse meausre based on the cumulative human exposure is developed to calculate the negative impact of traffic emissions. Regarding the negative impact as another objective function, the aforementioned formulations for dynamic second-best congestion pricing are further expanded to integrate the computation procedure of the negative impact, thus to finally obtain the multi-objective formulations for dynamic second-best congestion pricing considering environmental externality.. A dual-based heuristic algorithm is proposed to solve the formulations containing a large number of integer variables, which sequentially solves some relaxed nonlinear (sub)problem and some bi-objective knapsack updating problem. The cell-based dynamic traffic assignment algorithm is employed as a subroutine to obtain the traffic flow information and the objective values, every time before the execution of the relaxed nonlinear (sub)problem. Numerical examples based on various road networks of different patterns will be carried out to demonstrate the proposed models and algorithm. .

本课题研究考虑环境效应的动态次优拥挤收费问题的建模及求解方法。首先在动态用户平衡条件下为三种主要形式的次优拥挤收费（即边界线收费、区域收费和高乘坐率车道收费）的收费方案包括收费位置分别建立不同的约束条件，得到带平衡约束的混合整数规划模型。结合基于细胞模型的动态交通分配算法，开发兼顾计算量和计算精度的尾气排放量计算模型和非稳态尾气扩散模型，考虑气象条件对污染物扩散的影响，建立风险规避型的人口暴露指标来衡量尾气排放对环境的负面效应。将环境效应作为另一目标函数，并将其计算过程作为约束条件，对动态次优拥挤收费模型进行扩展，得到考虑环境效应的动态次优拥挤收费模型。所得模型为多目标且带平衡约束的大规模混合整数规划问题，本课题将开发基于对偶变量的启发式算法，通过求解一系列松弛子问题和多目标更新问题来求最优解。最后，课题将结合多个不同形式的典型城市路网，通过数值实验来验证所建模型和算法的有效性。

考虑环境效应的次优拥挤收费研究尚不完善，一方面网络化建模的研究对象局限于边界线收费，且求解算法对计算机资源的要求较高，另一方面对车辆排放环境负效应的计算方法比较宏观，无法反映动态变化的交通状况对尾气排放的影响。因此本项目的研究目标是为考虑环境效应的动态次优拥挤收费问题开发一套系统的建模方法，并设计有效可行的求解算法。. 本研究首先采用统一的建模方法为三种主要形式的次优拥挤收费（即边界线收费、区域收费和高乘坐率车道收费）的收费方案，包括收费位置和收费额度等，建立了相似的约束条件，所得均为混合整数的线性约束集。由于高乘坐率车道收费问题必须区分高乘坐率车辆和单乘坐率车辆，本研究进一步考虑多交通模式共存的情况（主要包括高乘坐率车辆出行、单乘坐率车辆出行、公共交通出行），建立了多模式交通平衡模型。并且基于对偶理论，将该模型转化为一组非线性约束。. 动态交通分配的数值求解算法中有一类依赖于细胞模型来传播交通流并计算实时出行时间。每个细胞携带动态的交通信息，使路网内交通流的分辨率得到提高。结合这一点，本研究开发了基于细胞模型和车辆驾驶模式的尾气排放量计算模型，以及基于细胞模型的尾气污染物扩散模型，从而获得了每一时刻尾气污染物的空间分布情况。结合人口分布情况及气象条件的不确定性因素，本研究采用了鲁棒的风险规避型人口暴露指标来衡量交通排放对环境的负面影响。. 本研究最终建立了以环境负效应和系统旅行时间最小化为双目标的动态次优拥挤收费模型，这些模型都是带平衡约束的大规模非线性混合整数规划问题。项目组开发了基于对偶变量的启发式算法来进行求解，该算法包括求解一系列松弛子问题和多目标更新问题，可同时优化收费位置和收费额度等，所得次优拥挤收费方案可直接用于实践。最后，项目组通过构建多个不同规模的数值实验验证了所建立的各约束条件、仿真模型及优化算法的有效性。