【作者】 冮龙晖；

【导师】 姜桂艳；

【作者基本信息】 吉林大学 ， 交通信息工程及控制， 2007， 博士

【摘要】 城市道路交通拥挤成为困扰世界各大城市的主要社会问题之一,严重影响着城市的可持续发展和人们的日常工作与生活。快速、准确地发现路网中发生的交通拥挤,并估计出拥挤未来一段时间内的扩散范围和持续时间,对于制定合理有效的交通拥挤疏导策略具有重要意义,本文对城市道路交通状态判别及拥挤扩散范围估计方法进行了深入研究。本文的主要研究工作及创新之处体现在:对基于GPS浮动车的交通信息采集的时空间隔确定方法和浮动车样本量的确定方法进行了研究,并使用模拟数据和实测数据进行了实证分析;以单元传输模型(Cell Transmission Model,CTM)为基础,设计了用于城市道路的交通参数在线估计方法,并将卡尔曼滤波方法嵌入到CTM中以提高交通参数的估计精度,最后使用模拟数据对算法的有效性进行验证,并对两种算法进行了对比分析;设计了基于时间序列相似性搜索的交通参数动态多步预测方法,对预测步数、初始时间序列长度、时间序列相似性搜索时长等参数的确定进行了讨论,最后使用实测数据来验证算法的有效性;给出了度量交通拥挤的绝对标准和相对标准的概念,分别设计了基于GPS浮动车数据和固定检测器数据的拥挤度量标准和交通状态判别方法,并分析了浮动车样本量不足时交通状态判别结果的可靠性,最后使用模拟数据进行案例分析;提出了交通拥挤扩散范围估计方法,包括单个路段交通拥挤的扩散范围估计、路网交通拥挤的扩散范围估计和交通拥挤持续时间的估计,最后使用模拟数据对方法进行了验证。论文中涉及到的研究内容、研究方法和研究结论是对城市道路交通状态判别及拥挤扩散范围估计方法的一个探索,可以为城市道路交通拥挤疏导策略的制定提供一定的理论依据和工程参考。更多还原

【Abstract】 Traffic congestion has been a global problem and makes serious affects on the persistence development of city and living of human beings. It is significant for decision-making of congetion dispersal to find congestion rapidly and estimate congestion propagation trend reliably. This dissertation performed systematic studies on the method of traffic state identification and congestion propagation estimation for urban road. The main research works conclude:(1) Study on the Data Collecting Technology based on GPS Probe VehiclesThe GPS probe vehicles have been a promising technology for traffic information collection detection with the GPS technology widely applied in transportation field. Some parameters settings of the GPS probe vehicles in data collecting will severe affect accuracy and reliability of the traffic information collection, mainly including link-dividing, sampling time interval of GPS data, transmission interval of GPS data, analysis interval of traffic data and probe vehicles sample-size. Considering the accuracy and cost of taffic data colletion, real-time and reliability of traffic state identification, this paper studies the method of special and temporal interval optimizing and the method of the GPS probe vehicles sample size.(2) Study on the Method of Traffic Papameters On-Line EstimationTraffic parameter data is crucial and necessary for traffic state survilliance and control systems. But due to the limitation of cost and actual condition, it is impossible to fix traffic dectectors at any spots, so that it is difficult to obtain the traffic paramters data of all spots. So it is significant important of the traffic parameters on-line estimation. This dissertation adopted Cell Transmission Model design a traffic parameter estimation method, inclue traffic flow, density, occupation and queue length. In order to improve the estimation acuuracy, Kalman Filter Technology was applied into CTM. Analysis results of simulated data showed that the proposed method can be used to on-line estimate traffic parameters of urban road.(3) Study on the Method of Traffic Parameters Dynamic Multi-steps PredictionThe becoming tendency information of traffic parameters is important for the making of traffic management decision and travel decision. The method of dynamic multi-steps traffic parameters prediction was proposed based on similarity searching technology of time series and common prediction method. Then some critical problems were analyzed and discussed, including the prediction steps, length of priminary series and time to searching similar time series. At last, the case study was performed with actual data, the results showed that the proposed multi-steps predition method have nice practicability.(4) Study on the Method of Traffic State Automatic IdentificationIndentifying the traffic abnormal state promptly and accurately is very important for advance the efficiency and effect of traffic manage. First the thesis defines absolute criterion and relative criterion for traffic congestion measure. Second, designed a traffic state indentification method based on GPS data and fixed detector data respectively. And then analysis the affect of probe vehicles sample-size on the reliability of traffic state identification result. Last the simulated data was used to validate the method proposed.(5) Study on the Estimation Method of Traffic Congestion PropagationThe information of the space propagation trend and duration of the traffic congestion are very important for the traffic management to make reasonable and effective measures for disperse congestion. This thesis analyzed the rule of traffic congestion propagation, and put forward the method of traffic congestion propagation estimation for single link and road network, and congestion duration estimation method. Finally, simulated data was used to test the proposed methods.更多还原