基于Kalman滤波的高速公路交通流实时状态估计方法研究

发布时间：2018-12-31 21:13

【摘要】：近年来,随着社会经济水平不断的提高,我国高速公路有了迅猛的发展,在社会经济活动中扮演着重要的运输作用,但是频发的高速公路交通事故也给社会经济带来巨大的经济损失。快速、准确地掌握和估计高速公路交通流状态,对于制定合理、有效的高速公路管控策略具有重要的意义,有利于缓解高速公路交通拥堵和提高高速公路安全性。然而,由于有限的检测设备无法提供高速公路全方位交通运行状态,因此,本文依托国家自然科学基金项目“高速公路车速离散特征、机理及控制方法研究”和南京市科技局项目“高速公路交通流状态估计与安全预警系统”,从实测数据出发,深入研究高速公路连续多断面交通流参数的时空关联特性,通过建立基于卡尔曼滤波的高速公路交通流实时状态估计方法,实时估计路段检测盲区的交通流状态,为制定高效的高速公路管控策略提供理论基础和技术支撑。首先,从实测数据出发,在统计学层面上引入交通流参数时空相关系数,分析高速公路多断面连续检测器交通流参数的时间相关性和空间相关性,为下文检测盲区的交通流状态估计提供数据基础,同时确定下文模型状态估计的路段区域长度。其次,研究不同宏观交通流模型实际估计效果,通过遗传算法对不同宏观交通流模型参数进行在线标定,并将标定后的模型应用于时空关联性较强的断面进行状态估计,以统计学评价指标最优为目标,选取交通流状态估计的最佳交通流模型；同时对模型的参数进行了敏感性分析。进一步对模型精度与检测间隔和路段距离的关系进行探讨。最终选取Jiang-Zhu-Wu模型作为交通流状态估计模型,其中自由流速度和阻塞传播速度为模型关键参数,模型在检测间隔为30s、路段划分距离为800m时状态估计效果最优。再次,以Kalman滤波“递推-估计-修正”原理为基础,分别构建了基于扩展Kalman滤波和基于无迹Kalman滤波的高速公路交通流状态估计模型,并给出状态估计的步骤。最后,基于实测数据对所构建的的交通流状态估计模型进行实例应用并评价其效果,主要包括两种状态估计模型对交通流状态突变的追踪能力、状态估计误差的对比分析等。同时亦对不同检测器布设方案下交通流状态估计模型的实施效果进行探讨,给出不同布设方案的误差,为检测器布设提供参考依据。
[Abstract]:In recent years, with the continuous improvement of social and economic level, the highway in China has a rapid development, which plays an important role in the social and economic activities. However, frequent highway traffic accidents also bring huge economic losses to the social economy. To grasp and estimate the state of expressway traffic flow quickly and accurately is of great significance for making reasonable and effective expressway management and control strategy, which is helpful to alleviate the traffic congestion and improve the safety of expressway. However, due to the limited detection equipment can not provide the highway traffic operation state, so this paper relies on the National Natural Science Foundation project "Highway speed discrete characteristics," The research on mechanism and control method "and Nanjing Science and Technology Bureau project" Expressway Traffic flow State estimation and Safety early warning system ", based on the measured data, the spatial-temporal correlation characteristics of continuous multi-section traffic flow parameters of expressway are deeply studied. By establishing the real-time state estimation method of freeway traffic flow based on Kalman filter, the real-time estimation of the traffic flow in the blind area of highway detection is carried out in real time, which provides the theoretical basis and technical support for the formulation of efficient expressway management and control strategy. Firstly, based on the measured data, the spatial-temporal correlation coefficient of traffic flow parameters is introduced to analyze the temporal and spatial correlation of traffic flow parameters of multi-section continuous detector in freeway. It provides a data basis for the traffic flow state estimation in the following blind areas, and determines the length of the road segment estimated by the following model state. Secondly, the actual estimation effect of different macroscopic traffic flow models is studied. The parameters of different macroscopic traffic flow models are calibrated online by genetic algorithm, and the calibrated models are applied to estimate the state of sections with strong temporal and spatial correlation. The optimal traffic flow model of traffic flow state estimation is selected. At the same time, the sensitivity of the model parameters is analyzed. The relationship between the accuracy of the model and the detection interval and the distance between the sections is discussed. Finally, the Jiang-Zhu-Wu model is selected as the traffic flow state estimation model, in which the free flow velocity and the congestion propagation velocity are the key parameters of the model. The model has the best effect when the detection interval is 30s and the partition distance is 800m. Thirdly, based on the principle of "Recursion-Estimator-Correction" of Kalman filter, the traffic flow state estimation models based on extended Kalman filter and unscented Kalman filter are constructed, and the steps of state estimation are given. Finally, the traffic flow state estimation model is applied to the traffic flow estimation model based on the measured data and its effect is evaluated, including the tracking ability of the two state estimation models to the sudden change of the traffic flow state, the comparative analysis of the state estimation error, and so on. At the same time, the effect of traffic flow state estimation model under different detector layout schemes is discussed, the error of different layout schemes is given, and the reference basis for detector layout is provided.
【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491.112

【相似文献】