基于扩展Logit的交通分配模型与算法研究
发布时间：2019-07-09 19:01
【摘要】：随着我国城市交通供需矛盾的日趋严重,交通问题已经成为制约城市发展的主要瓶颈之一。为了解决交通问题,改善交通状况,各级政府部门从交通供给和需求两方面同时出发,不断加大交通基础设施建设力度和提高交通需求管理水平。要保证交通规划和管理措施在一定程度上取得最佳效果,就需要采用先进的交通规划理论和方法来辅助决策。交通分配模型是现代交通规划理论研究的重要内容之一,也是城市道路交通网络设计和拥挤道路收费研究中的一个核心技术。在此背景下,本论文主要针对城市交通网络,研究能够更准确描述出行者路径选择行为的交通分配模型,并将其应用于交通网络设计和拥挤道路收费的研究中,使本论文研究形成一个较为完整的体系,以期为交通管理和规划者的决策提供一定的支持。本文主要完成了以下几方面的研究： (1)为了克服基于传统Logit的交通分配模型所存在的不相关备选方案的独立性(ⅡA),采用考虑路径重叠的路径感知Logit (RPL-O)模型和路径尺度修正Logit (PSCL)模型,分别构建了基于RPL-O和PSCL的随机用户平衡模型,并证明了模型的等价性以及解的唯一性；设计了求解所提出两种新模型的基于路径的相继平均法(MSA)、连续权重平均法(MSWA)以及自动调整平均法(SRA)；通过数值实验对比了新模型与传统基于Logit的随机用户平衡模型的交通流分配结果,分析了三种算法的求解效率,结果表明SRA算法是本文所采用的求解算法中效率最高的算法。 (2)阐述了C-Logit、路径尺度Logit、广义巢式Logit、配对组合Logit路径选择模型的选择概率形式及模型参数的确定方法；分别构建了基于C-Logit、路径尺度Logit、RPL-O.路径尺度修正Logit、广义巢式Logit、配对组合Logit模型的多用户多准则随机用户平衡模型,并对所提出的六种新模型的等价性以及解的唯一性进行了证明；对比分析了模型中路网结构的表示方法和路径选择概率的计算过程；设计了通用的求解所提出模型的基于路径的SRA算法；通过数值实验对比了新模型与基于Logit的多用户多准则随机用户平衡模型的交通分配结果,分析了模型参数变化对分配流量的影响。 (3)分析了已有连续网络设计研究存在的不足,以出行总阻抗与总投资额之和最小为上层目标函数,以本文第3章所构建的两种模型以及另外四种基于扩展Logit的随机用户平衡模型为下层模型,构建出六种新的连续交通网络设计模型；设计了将遗传算法与SRA算法相结合的组合算法；通过数值实验验证了本文算法的有效性,分析了下层采用基于扩展Logit的随机用户平衡模型与传统确定性用户平衡模型所得到网络设计方案的区别,分析了模型参数变化时,网络设计结果的变化情况。 (4)通过分析传统次优拥挤收费研究中存在的不足,说明考虑多用户多准则以及采用基于扩展Logit的交通分配模型来研究拥挤收费的意义；以系统出行总时间最小为上层目标函数,以基于扩展Logit的多用户多准则随机用户平衡模型为下层模型,构建出六种新的次优拥挤收费模型；设计了将模拟退火算法与SRA算法相结合的组合算法；通过数值实验检验了算法的有效性,对比了下层采用基于扩展Logit的多用户多准则随机用户平衡模型与多用户多准则确定性用户平衡模型所得到的路段收费方案的区别,分析了模型参数变化时,拥挤收费方案的变化情况。
文内图片：
图片说明：不同分散系数0卜的RMSE
[Abstract]:With the increasing supply and demand of urban traffic, the problem of traffic has become one of the main bottlenecks that restrict the development of the city. In ord to solve that traffic problem and improve the traffic condition, the government departments at all levels, from the two aspects of transportation supply and demand, continuously increase the construction of the transportation infrastructure and improve the management level of the traffic demand. To ensure that traffic planning and management measures achieve the best effect to some extent, it is necessary to adopt advanced traffic planning theory and method to assist in decision-making. The traffic distribution model is one of the important content of the modern traffic planning theory, and it is also a core technology in the study of urban road traffic network design and congestion road charging. In this background, the paper mainly focuses on the urban traffic network, which can describe the traffic distribution model of the pedestrian's path selection behavior more accurately and apply it to the study of the traffic network design and the congestion road charging, so that this paper will form a more complete system. In order to provide some support for the decision-making of traffic management and planners. The following aspects are mainly completed in this paper: (1) In order to overcome the independence (II A) of the unrelated alternatives existing in the traffic distribution model based on the traditional Logit, the path-aware Loit (RPL-O) model and the path-scale modified Loit (PSCL) mode considering the path overlap are adopted. The random user equilibrium model based on RPL-O and PSCL is constructed, and the equivalence of the model and the uniqueness of the solution are proved. The path-based successive average method (MSA), the continuous weight average method (MSWA) and the automatic adjustment average method (SRA) are designed to solve the two new models proposed. In this paper, the traffic flow distribution results of the new model and the traditional Logit-based random user equilibrium model are compared by numerical experiments, and the efficiency of the three algorithms is analyzed. The results show that the SRA algorithm is the most efficient algorithm in the algorithm used in this paper. (2) The selection probability and model parameters of C-Logit, path scale Logit, generalized nested Logit, paired combination Logit path selection model are set forth, and the C-Loit, path scale Logit, RP, respectively, are constructed. L-O. Path-scale modified Loit, generalized nested Logit, the multi-user multi-criterion random user equilibrium model of the paired combination Logit model, and the equivalence of the six new models and the uniqueness of the solution. The paper proves that the method of representation of the road network structure in the model and the calculation process of the route selection probability are compared and analyzed, and the path-based SR for the model proposed by the general solution is designed. A. The traffic distribution results of the new model and the multi-user multi-criteria random user equilibrium model based on Logit are compared by numerical experiments, and the distribution flow of the model parameter change is analyzed. (3) The deficiency of the existing continuous network design research is analyzed, and the sum of total impedance and total investment is the upper target function. The two models constructed in Chapter 3 of this paper and four other random user equilibrium models based on extended Logit are analyzed. In this paper, six new design models of continuous traffic network are built for the lower layer model, and a combination algorithm combining the genetic algorithm with the SRA algorithm is designed. The effectiveness of the method is analyzed, and the difference between the random user equilibrium model based on extended Logit and the network design scheme of the traditional deterministic user equilibrium model is analyzed, and the network design result is analyzed when the model parameters are changed. (4) To study the significance of multi-user multi-criteria and to use the extended Logit-based traffic distribution model to study the significance of the congestion charge by analyzing the shortcomings existing in the traditional sub-optimal congestion charging study, and to minimize the total travel time of the system In this paper, six new sub-optimal congestion charging models are constructed based on the extended Logit-based multi-user multi-criteria random user equilibrium model, and a combination algorithm combining the simulated annealing algorithm with the SRA algorithm is designed. The effectiveness of the algorithm is compared, and the difference between the multi-user multi-criterion random user equilibrium model and the multi-user multi-criterion deterministic user equilibrium model based on the extended Logit is compared, and the congestion charge is analyzed when the model parameters are changed
【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：U491
本文编号：2512372
文内图片：
图片说明：不同分散系数0卜的RMSE
[Abstract]:With the increasing supply and demand of urban traffic, the problem of traffic has become one of the main bottlenecks that restrict the development of the city. In ord to solve that traffic problem and improve the traffic condition, the government departments at all levels, from the two aspects of transportation supply and demand, continuously increase the construction of the transportation infrastructure and improve the management level of the traffic demand. To ensure that traffic planning and management measures achieve the best effect to some extent, it is necessary to adopt advanced traffic planning theory and method to assist in decision-making. The traffic distribution model is one of the important content of the modern traffic planning theory, and it is also a core technology in the study of urban road traffic network design and congestion road charging. In this background, the paper mainly focuses on the urban traffic network, which can describe the traffic distribution model of the pedestrian's path selection behavior more accurately and apply it to the study of the traffic network design and the congestion road charging, so that this paper will form a more complete system. In order to provide some support for the decision-making of traffic management and planners. The following aspects are mainly completed in this paper: (1) In order to overcome the independence (II A) of the unrelated alternatives existing in the traffic distribution model based on the traditional Logit, the path-aware Loit (RPL-O) model and the path-scale modified Loit (PSCL) mode considering the path overlap are adopted. The random user equilibrium model based on RPL-O and PSCL is constructed, and the equivalence of the model and the uniqueness of the solution are proved. The path-based successive average method (MSA), the continuous weight average method (MSWA) and the automatic adjustment average method (SRA) are designed to solve the two new models proposed. In this paper, the traffic flow distribution results of the new model and the traditional Logit-based random user equilibrium model are compared by numerical experiments, and the efficiency of the three algorithms is analyzed. The results show that the SRA algorithm is the most efficient algorithm in the algorithm used in this paper. (2) The selection probability and model parameters of C-Logit, path scale Logit, generalized nested Logit, paired combination Logit path selection model are set forth, and the C-Loit, path scale Logit, RP, respectively, are constructed. L-O. Path-scale modified Loit, generalized nested Logit, the multi-user multi-criterion random user equilibrium model of the paired combination Logit model, and the equivalence of the six new models and the uniqueness of the solution. The paper proves that the method of representation of the road network structure in the model and the calculation process of the route selection probability are compared and analyzed, and the path-based SR for the model proposed by the general solution is designed. A. The traffic distribution results of the new model and the multi-user multi-criteria random user equilibrium model based on Logit are compared by numerical experiments, and the distribution flow of the model parameter change is analyzed. (3) The deficiency of the existing continuous network design research is analyzed, and the sum of total impedance and total investment is the upper target function. The two models constructed in Chapter 3 of this paper and four other random user equilibrium models based on extended Logit are analyzed. In this paper, six new design models of continuous traffic network are built for the lower layer model, and a combination algorithm combining the genetic algorithm with the SRA algorithm is designed. The effectiveness of the method is analyzed, and the difference between the random user equilibrium model based on extended Logit and the network design scheme of the traditional deterministic user equilibrium model is analyzed, and the network design result is analyzed when the model parameters are changed. (4) To study the significance of multi-user multi-criteria and to use the extended Logit-based traffic distribution model to study the significance of the congestion charge by analyzing the shortcomings existing in the traditional sub-optimal congestion charging study, and to minimize the total travel time of the system In this paper, six new sub-optimal congestion charging models are constructed based on the extended Logit-based multi-user multi-criteria random user equilibrium model, and a combination algorithm combining the simulated annealing algorithm with the SRA algorithm is designed. The effectiveness of the algorithm is compared, and the difference between the multi-user multi-criterion random user equilibrium model and the multi-user multi-criterion deterministic user equilibrium model based on the extended Logit is compared, and the congestion charge is analyzed when the model parameters are changed
【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：U491
【参考文献】
相关期刊论文 前10条
1 蔡金,高自友;求解城市交通连续平衡网络设计问题的混合算法[J];北方交通大学学报;2002年02期
2 宋一凡,高自友;求解连续平衡网络设计问题近似解的启发式算法[J];北方交通大学学报;1998年06期
3 王力,王川久,沈晓蓉,范跃祖;基于模糊旅行时间的动态交通分配模型[J];北京航空航天大学学报;2005年10期
4 王卫杰;王炜;;基于配对组合logit模型的路线选择[J];东南大学学报(自然科学版);2010年04期
5 邱松林;程琳;许项东;;基于路径长度的Logit型随机用户均衡模型[J];东南大学学报(自然科学版);2012年01期
6 杨清华,贺国光,马寿峰;对动态交通分配的反思[J];系统工程;2000年01期
7 徐兵;朱道立;;多用户多准则固定需求随机交通均衡变分模型[J];公路交通科技;2007年04期
8 何胜学;;基于出行时间预算的多用户交通分配模型及算法[J];公路交通科技;2009年05期
9 龚峻峰;余志;何兆成;;一种基于路段惩罚法的合理路径集生成算法[J];公路交通科技;2009年09期
10 熊伟;严新平;;一种多模式下考虑排放的交通分配模型及其算法研究[J];公路交通科技;2009年09期
，本文编号：2512372
本文链接：https://www.wllwen.com/kejilunwen/jiaotonggongchenglunwen/2512372.html
