高维多目标离散交通网络设计模型与算法
发布时间：2019-02-23 19:08
【摘要】：城市机动化的快速发展,使交通需求与交通供给之间的矛盾不断扩大,从而产生了诸如交通拥堵、交通环境污染、交通事故频发等交通问题。交通规划是一种提供交通设施的方案,若能编制科学合理的交通规划,可以从源头上避免或减少交通问题的发生。然而,常用的交通规划编制方法是一种定性的方法,缺乏对实际网络用户出行行为的考虑,这种规划方案实施之后,常会出现交通改善不明显甚至交通问题加剧的现象,同时也浪费了有限的投入资金。因此,在有限的投入资金情况下,在考虑网络用户出行行为的基础上,定量地研究交通网络设计问题有着重要的应用价值。本文首先深入分析了高维多目标交通网络设计问题的相关基础理论。然后在详述双层规划模型、多目标优化模型及典型交通网络设计模型的基础上,从政府多部门决策和用户出行行为的角度出发,构建了同时优化系统阻抗、尾气排放、投资费用、饱和度和用户出行行为目标的离散交通网络设计双层规划模型。其中,上层规划模型将多个优化目标化成多个单目标建模,路段通行能力增量约束据规划通行能力设置；下层规划模型采用固定需求下的用户平衡配流模型。随后,为了对构建的模型进行有效求解,从保证求解算法良好的收敛性和分布性的角度出发,在第2代非劣排序遗传算法(Non-dominated Sorting Genetic Algorithm Ⅱ,简称NSGA-Ⅱ)的基础上,基于第k小距离双层截断策略做了改进,设计了改进的NSGA-Ⅱ,并采用Matlab语言开发了相应的算法程序。最后,以Nguyen-Dupuis网络作为算例,通过对比分析NSGA-Ⅱ和改进的NSGA-Ⅱ对三维、四维优化模型的求解结果,证明了改进的NSGA-Ⅱ可以有效求解基本的高维多目标离散交通网络设计模型,并结合相关专业知识,得到了8个交通网络设计优化方案。
[Abstract]:With the rapid development of urban motorization, the contradiction between traffic demand and traffic supply has been enlarged, resulting in traffic problems such as traffic congestion, traffic environment pollution, frequent traffic accidents and so on. Traffic planning is a scheme to provide transportation facilities. If we can draw up scientific and reasonable traffic planning, we can avoid or reduce the occurrence of traffic problems at the source. However, the commonly used traffic planning method is a qualitative method, which lacks consideration of the travel behavior of actual network users. After the implementation of this planning scheme, traffic improvement is not obvious and even traffic problems are aggravated. At the same time, it is a waste of limited investment. Therefore, under the condition of limited investment and taking into account the travel behavior of network users, quantitative research on traffic network design has important application value. In this paper, the basic theory of high-dimensional multi-objective traffic network design is analyzed. Then, on the basis of describing the bilevel programming model, multi-objective optimization model and typical traffic network design model, from the angle of government multi-department decision-making and user travel behavior, the system impedance and exhaust emission are optimized at the same time. A two-level programming model for discrete traffic network design with investment cost saturation and user travel behavior objectives. Among them, the upper planning model transforms multiple optimization objectives into multiple single-objective models, and the incremental capacity of road sections is constrained according to the planned capacity setting; the lower level programming model adopts the user equilibrium assignment model under fixed demand. Then, in order to solve the model effectively, in order to ensure the good convergence and distribution of the algorithm, based on the second generation of non-inferior sorting genetic algorithm (NSGA- 鈪，
本文编号：2429112
[Abstract]:With the rapid development of urban motorization, the contradiction between traffic demand and traffic supply has been enlarged, resulting in traffic problems such as traffic congestion, traffic environment pollution, frequent traffic accidents and so on. Traffic planning is a scheme to provide transportation facilities. If we can draw up scientific and reasonable traffic planning, we can avoid or reduce the occurrence of traffic problems at the source. However, the commonly used traffic planning method is a qualitative method, which lacks consideration of the travel behavior of actual network users. After the implementation of this planning scheme, traffic improvement is not obvious and even traffic problems are aggravated. At the same time, it is a waste of limited investment. Therefore, under the condition of limited investment and taking into account the travel behavior of network users, quantitative research on traffic network design has important application value. In this paper, the basic theory of high-dimensional multi-objective traffic network design is analyzed. Then, on the basis of describing the bilevel programming model, multi-objective optimization model and typical traffic network design model, from the angle of government multi-department decision-making and user travel behavior, the system impedance and exhaust emission are optimized at the same time. A two-level programming model for discrete traffic network design with investment cost saturation and user travel behavior objectives. Among them, the upper planning model transforms multiple optimization objectives into multiple single-objective models, and the incremental capacity of road sections is constrained according to the planned capacity setting; the lower level programming model adopts the user equilibrium assignment model under fixed demand. Then, in order to solve the model effectively, in order to ensure the good convergence and distribution of the algorithm, based on the second generation of non-inferior sorting genetic algorithm (NSGA- 鈪，
本文编号：2429112
本文链接：https://www.wllwen.com/kejilunwen/jiaotonggongchenglunwen/2429112.html
