基于有权网络子区划分的区域交通协调控制研究

发布时间：2018-02-10 01:12

本文关键词： 交通信号复杂网络有权划分协调控制多目标优化　出处：《南京邮电大学》2017年硕士论文　论文类型：学位论文

【摘要】：城市交通信号控制是缓解路网交通拥堵,提高路网资源利用率的主要手段。合理有效的交通信号控制策略,不仅需要研究路网中单路口的信号控制,还需要综合考虑整个路网的总体通行效率。由于交通路网过于复杂和庞大,直接对整个路网进行研究非常困难。将交通路网划分成若干个合理的信号控制子区,然后对每个子区协调控制是有效的解决方案。本文针对现有复杂路网子区划分的不足,提出一种合理、正确的划分方法。然后基于各划分子区,构建子区间协调控制的多目标优化模型。最后对NSGA-II算法进行改进,求解子区协调控制的信号配时。主要研究工作如下:(1)针对现有复杂网络社团划分算法难以有效进行复杂路网的子区划分,提出一种改进的Newman社团划分算法,能有效地对复杂交通路网进行子区划分。本文选用交通路网中的两路口间的距离和单位时间的车流量构建边权,再结合复杂网络的度特性构建点权,重新定义模块度的概念。由于交通流是一个动态值,可以选取不同时间段的交通流,实现区域路网的动态划分。测试实验表明,改进后的算法能够敏锐的辨识路段间的耦合程度,实现对区域路网合理、正确的划分。(2)针对现有区域交通协调控制的不足,基于上述子区划分构建了子区间协调控制的多目标优化模型。根据划分子区的特点,以各子区的信号周期为优化变量,以区域交通路网的平均延误时间及下游子区路口的平均停车率为优化指标,构建了子区间协调控制的多目标优化模型,克服了现有区域交通协调控制优化目标单一、运算复杂或者在运算前需要较多先验知识的不足。(3)针对NSGA-II算法在求解子区协调控制的信号配时时出现解分布的不均匀,种群的多样性较弱等问题,本文在原有变异算子的基础上提出一种基于泊松随机数的变异算子,并在生成初始种群的时候,先对种群进行筛选,保证优质的个体被保留。仿真测试结果表明,改进后的算法对Pareto前沿面具有较强的探索能力,并且能够保持种群的多样性。用改进后的算法求解子区协调控制的信号配时,使得各子区组成的交通路网平均延误时间以及停车率都得到了改善。
[Abstract]:Urban traffic signal control is the main means to alleviate traffic congestion and improve the utilization of road network resources. A reasonable and effective traffic signal control strategy not only needs to study the signal control of a single intersection in the road network. It is also necessary to take into account the overall traffic efficiency of the whole network. Because the traffic network is too complex and large, it is very difficult to study the whole network directly. The traffic network is divided into several reasonable signal control sub-areas. Then the coordinated control for each sub-area is an effective solution. In this paper, a reasonable and correct method is proposed for subdivision of complex road network, and then, based on the subdivision of each sub-area, a reasonable and correct method is proposed. The multi-objective optimization model of subinterval coordinated control is constructed. Finally, the NSGA-II algorithm is improved. The main research work is as follows: (1) aiming at the difficulty of existing complex network community partition algorithms, an improved Newman community partition algorithm is proposed. In this paper, the distance between two intersections and the traffic flow per unit time in the traffic network are selected to construct the side weight, and then the point weight is constructed by combining the degree characteristics of the complex network. The concept of modular degree is redefined. Because traffic flow is a dynamic value, traffic flow in different time periods can be selected to realize the dynamic division of regional road network. The improved algorithm can acutely identify the coupling degree between road sections, realize the rational division of the regional road network, and correct division. 2) aiming at the shortcomings of the existing regional traffic coordination control, The multi-objective optimization model of subinterval coordination control is constructed based on the above subarea partition. According to the characteristics of the subregion, the signal period of each sub-region is taken as the optimization variable. Taking the average delay time of the regional traffic network and the average stop rate of the intersection of the regional traffic network as the optimization indexes, the multi-objective optimization model of the sub-interval coordinated control is constructed, which overcomes the single optimization goal of the existing regional traffic coordination control. In view of the problem that the NSGA-II algorithm can solve the problems of uneven distribution of solution and weak diversity of population in solving the signal matching of subzone coordination control, the algorithm is complex in operation or requires more prior knowledge before operation. Based on the original mutation operator, a mutation operator based on Poisson random number is proposed in this paper. When the initial population is generated, the population is first screened to ensure that the high quality individual is retained. The simulation results show that, The improved algorithm has a strong ability to explore the front mask of Pareto and can keep the diversity of population. The improved algorithm is used to solve the signal timing of subregion coordinated control. The average delay time and parking rate of traffic network are improved.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491.54;O157.5

【相似文献】