城市路网动态转向建模与优先选择研究  

Network Representation and Sub-optimal Choice of Linkbased Dynamic Routing in Urban Network

作　　者：徐广宁[1] 王印海[1,2] 曾自强[2,3] 

机构地区：[1]哈尔滨工业大学交通科学与工程学院,哈尔滨150090 [2]华盛顿大学土木与环境工程学院,美国西雅图98195 [3]四川大学商学院不确定决策实验室,成都610065

出　　处：《交通运输系统工程与信息》2017年第4期132-137,共6页Journal of Transportation Systems Engineering and Information Technology

基　　金：国家自然科学基金(51138003)~~

摘　　要：确立网络表述方式是在城市路网中建立路径规划系统的基础工作之一.在不牺牲计算效率的前提下,网络表述方式既要能灵活反映出车辆在交叉口中所能进行的转向动作,又必须保证这些转向动作不会造成交叉口的安全隐患与通行效率的下降.针对这一需求,提出了一套面向决策点的网络表述方法,并在其数据结构的基础上,通过扩展现有的Dijkstra最短路搜索算法,在不改变以最短路为优化目标的前提下,以次优选择方式实现了干道优先和非左转优先的转向选择.以现实交通网络为模型,通过一系列实验验证算法在静态路径规划中的有效性.结果表明,算法以提高0.5%额外总出行成本的前提下,同时降低了11%的支道选择和21%的左转选择.Establishing network representation is usually the first step of implementing a route planning system in urban street networks. Without losing computation efficiency, network representation should be able to reflect all the possible vehicle routing movements that are allowed in city intersections considering transportation safety and mobility. To fulfill the requirement, a full package of decision-point oriented network representation is proposed in this paper. Based on the data structure of the network representation, two extensions on the conventional Dijkstra shortest path finding algorithm, not by changing its global optimality but using sub-optimal thresholds, are also proposed to establish priority turning for arterial first and left-turn avoiding principles. In order to validate the network representation and the extended algorithms, a series of static route planning experiments are conducted using the simulation model of a real-world street network. Test results manifest that 11%of branches and 21%of left-turn choices are avoided simultaneously only at the cost of 0.5%of extra total travel cost.

关 键 词：智能交通 路径规划 网络表示 动态转向 次优选择 

分 类 号：U492.2[交通运输工程—交通运输规划与管理]