基于检测器数据的道路网络交通运行状态可靠性在线分析与预测方法研究

Along with the increasing of dimension and complexity of road network and with the popularization of motor vehicle, congestion alleviating has been a highly complicated systems engineering which need the cooperation of many factors, many processes and many techniques. Intelligent Transportation Systems (ITS) is known as the best approach to enhance the efficiency and security of the existent road network. Especially, Advanced Traffic Management System (ATMS) and Advanced Traffic Information System (ATIS) in ITS are in the highest flight. .The traffic surveillance systems have been built in many cities in China which have played an important role in improving the efficiency and effectiveness of the dynamic decisions for traffic managers and drivers. But further upgrade could be done for the processing and utilizing of the detector data in these systems. One problem is the incapacity in screening out the sensitive links and intersections which would have strategic impact on the operation stability of the road network as a whole. Another problem is the lack of trend prediction for traffic state in some time intervals in the future..Under the conditions of road traffic congestion getting serious increasingly and of detector data online using insufficiently, our goal is to enhance the traffic operation stability of the road network by providing more valuable dynamic decision supporting information for traffic managers and drivers through digging the detector data online from a new perspective and pointing out the sensitive links and intersections which impact the network reliability a lot at present and short future. .Online methods of reliability analysis and prediction for traffic operation state on road network based on detector dada mainly include the following: (1) index system and their online measure methods of traffic operation state reliability for single link and intersection based on detector data; (2) online analysis methods of traffic operation state reliability for single path of a given OD, sub-network among a given OD, and regional network; (3) screening principles and methods of sensitive links and intersections based on the results of online analysis of network traffic operation state reliability; (4) predictability dynamic analysis method and multi-step predicting model for short-term traffic parameters based on detector data..The research achievements are remarkable in the aspects of academic significance and application possibility, because of the prospective improvement in the pertinency and predictability of dynamic decision supporting for traffic managers and drivers, in the formation preventing and evacuation accelerating of traffic congestion, and in the stability and efficiency of traffic operation on the road network.

针对目前道路交通拥堵日益严重的现实，本项目以提高路网交通运行稳定性为目标，通过从全新角度对各类交通检测器获取的实时交通数据进行更深层次的挖掘，指出当前和未来一段时间内影响路网交通运行可靠性的敏感单元，为交通管理者和交通出行者的动态决策提供更有价值的信息基础。研究内容主要包括：（1）基于检测器数据的路段/路口交通运行状态可靠性指标体系及其测度方法；（2）基于检测器数据的单一路径、OD 间子路网和区域路网的交通运行状态可靠性分析方法；（3）基于路网交通运行状态可靠性分析的敏感路段和路口确定方法；（4）基于检测器数据的短时交通参数动态可预测性在线分析方法和短时交通参数多步预测模型与算法。.本项目的研究成果对于改善交通管理者和交通出行者动态决策的预见性和有效性、预防交通拥堵形成和加快拥堵疏散、提升路网整体交通运行的稳定性和效率，具有重要的学术意义和广泛的应用前景。

在道路交通拥堵日益严重的背景下，本项目以提高路网交通运行稳定性为目标，通过对检测器获取的交通数据进行更深层次的挖掘，确定当前和未来一段时间内影响路网交通运行可靠性的敏感单元，可为交通管理者和交通出行者提供更为及时、可靠和有预见性的决策依据。.本项目的主要研究进展包括如下5个方面。.（1）提出了交通运行状态可靠性的定义，设计了包括长期可靠性（分时可靠度、全时可靠度）和现势可靠性（可靠性指数）的交通运行状态可靠性指标体系及其测度方法。.（2）针对课题组前期成果计算量大的问题，先后基于贝叶斯网络和因子定理提出了路网交通运行状态长期可靠性动态分析的改善方法；提出了路段交通运行状态长期可靠度和交通运行状态长期不可靠度指标及其计算方法，并以故障树分析为基础，设计了效率更高的路网交通运行状态长期可靠性在线分析方法；定义了交通运行状态现势可靠性指数的概念，确定了交通运行状态现势可靠性指数的等级划分标准，并分别设计了路网交通运行状态现势可靠性在线计算方法。.（3）以基于贝叶斯网络的交通运行状态长期可靠性分析结果为基础，提出了基于ABC分类的敏感路段和路口筛选方法；以基于故障树分析的交通运行状态长期可靠性分析结果为基础，针对交通数据获取程度的差异性，设计了影响路网交通运行状态长期可靠性的敏感路段和路口的确定方法；综合考虑路段/交叉口物理结构特性、承担交通量特性、交通运行特性和可靠性重要度特性，设计了基于多属性决策理论影响交通运行状态现势可靠性的敏感路段/交叉口的识别方法。.（4）提出了基于多时间尺度一步外推的短时交通参数多步预测方法、基于动态神经网络的交通参数迭代多步预测双层模型和基于k-NN的交通参数直接多步预测双层模型，进一步提高了短时交通参数动态可预测性在线分析与短时交通参数多步预测的效率和效果。.（5）针对预测管理和目标管理的需要，对路网交通运行状态可靠性的预测与分配方法进行了扩展研究，构建了基于粒子群优化小波神经网络的交通状态可靠性多步预测双层模型，设计了基于AHP的路网交通运行状态长期可靠性分配方法和基于AHP的交通状态现势可靠性分配方法。.本项目研究成果拓展了路网可靠性研究的范畴，对于改善交通管理者和交通出行者动态决策的预见性和有效性、预防交通拥堵形成和加快拥堵疏散、提升路网整体交通运行的稳定性和效率，具有重要的学术意义和广泛的应用前景。