具备自我更新机制和预测能力的实时动态交通诱导方法

发布时间：2018-04-12 14:37

本文选题：智能诱导方法 + 行程时间预测　；参考：《山东大学》2017年硕士论文

【摘要】：近几年随着我国经济的飞速发展,机动车数量始终保持较快的增长速度,机动车数量的大幅度增长导致城市交通拥堵问题日益严重。如何在现有道路基础设施的基础上快速、有效的解决交通问题已经成为智能交通研究领域一个亟待解决的问题,而动态路径诱导系统又是智能交通研究领域的关键所在。其中动态路径诱导系统的关键是如何充分利用现有道路资源将道路网中行驶的车辆合理的分配到不同的道路上,从而使整个路网运行状况达到最佳状态。本文提出了一种新型的实时车辆诱导方法,结合短时道路交通状况预测、动态交通分配、智能发布算法和两类实时监控自我调整模式,形成了一种能够动态自我验证和更新的智能诱导方法。相比传统交通诱导方法,本文提出的方法更能适应交通状况动态变化、驾驶员对发布信息的服从度不确定等实际情况,通过自我校验和自我更新的机制,能够更好的实现平衡交通压力的效果。针对智能诱导方法所需的短时交通预测,本文提出了一种基于BP神经网络和改进卡尔曼滤波的组合预测模型,并以济南市经十路真实数据为例,在四种场景下,从准确性和可靠性两个方面对比了 BP神经网络预测模型、改进卡尔曼滤波预测模型以及组合模型的预测效果,结果证明组合预测模型的有效性。针对智能诱导方法所需的动态交通分配,本文基于动态交通分配中的动态系统最优模型和动态用户最优模型,并从路网管理者和交通出行者之间的利益出发,运用博弈论分析两者之间的利益冲突,并提出了一种用遗传算法优化的博弈论分配方法来求解路网中的流量分配问题。选取山东大学千佛山校区区域路网进行案例分析,验证遗传算法优化博弈论分配方法的有效性。针对智能诱导方法的智能动态发布和实时校验,本文在分析驾驶员路径选择行为影响因素以及诱导信息发布策略影响因素的基础上,提出了一种诱导信息发布及校验机制,通过分析诱导信息发布之后的实际诱导效果与预期诱导效果之间的差异来判定是否需要调整发布策略或重新进行交通分配,从而形成一个集时间预测、交通流分配、诱导信息发布、诱导效果校验于一体的完整诱导环路。
[Abstract]:In recent years, with the rapid development of China's economy, the number of motor vehicles has always maintained a rapid growth rate, the number of motor vehicles increased greatly in the city traffic congestion is becoming a serious problem. Based on the existing road infrastructure quickly, solve the traffic problem effectively has become an urgent problem of intelligent traffic field, and dynamic route guidance system is the key to the field of intelligent transportation research. One of the key dynamic route guidance system is how to make full use of the existing road resource allocation will be traveling the road network of the vehicle to the right path, so that the entire network operating conditions to achieve the best state. This paper proposes a real-time vehicle model induction method the combination of short-term traffic prediction, dynamic traffic assignment, intelligent release self adjustment pattern algorithm and two kinds of real-time monitoring And form a dynamic self checking and updating of the intelligent induction method. Compared with the traditional traffic guidance method, this method can adapt to the dynamic change of traffic conditions, the driver information compliance uncertainty of the actual situation, through self check and self renewal mechanism, to achieve a better balance of traffic pressure the effect of short term traffic. For intelligent induction methods required for prediction, this paper proposes a method based on BP neural network and improved Calman filter combination forecasting model, and taking Ji'nan city road by the ten real data as an example, in the four scenario, from the two aspects of accuracy and reliability compared with the BP neural network forecasting model Calman, improved filtering prediction model and the prediction effect of the combination model, combination forecasting results prove the effectiveness of the model. According to the dynamic traffic assignment method for intelligent induction, In this paper, the dynamic system optimal model and the dynamic user optimal model based on dynamic traffic assignment, and from the point of network administrators and travelers between the interests of using game theory to analyze the conflict of interests between the two, and proposes a game based on genetic algorithm optimization method on the distribution of the flow distribution problem of the road network. Select the Qianfo Hill campus of the Shandong University regional network for case analysis, to verify the effectiveness of the genetic algorithm to optimize the distribution of game theory method. For the intelligent dynamic induction method of intelligent release and real-time calibration, based on the analysis of driver route choice behavior influencing factors and basic factors of the induced release of information strategy, put forward a kind of guidance and information and check mechanism the difference between the effect of induction, through the analysis of the actual information released after induced by induced effect and expected to determine is No need to adjust the publishing strategy or redistribute the traffic, so as to form a complete induction loop that integrates time prediction, traffic assignment, induced information release and induction effect verification.

【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495

【参考文献】