当前位置: 首页>博士论文>资源详情
高速公路路网交通运行状态关键问题的研究
中文摘要

交通运输业迅速发展及居民出行需求的大幅度增加,导致道路交通运输压力日益增加,尤其是承载着城市间互通及大城市周边的高速公路运输压力更加严重。一旦交通事件发生,会极大地影响道路的通畅程度甚至会引发连锁效应,对高速公路运输能力造成巨大的影响。因此,有必要对高速公路路网运行状态的关键问题进行深入的研究,以期来实现对高速公路运行状态的有效监管。针对这一问题,本文以在高速公路上采集到的浮动车轨迹数据为基础,从浮动车数据去噪、高速公路交通事件检测、高速公路路网交通运行状态量化这三个角度,层层递进地进行高速公路道路信息研究,具体研究内容如下: 浮动车数据的时空分布决定了高速公路路网运行状态研究的可行性及范围,虽然对大量浮动车进行监控,采集到了海量轨迹数据,但是由于车辆行驶轨迹具有很强的随机性导致数据在时空上分布不均,导致研究范围难以界定。针对这一问题,通过研究道路交通信息所需的数据量下限并结合本文数据时空分布情况的分析,确定高速公路道路的研究范围及道路信息的发布周期。 为了解决浮动车数据在采集过程中各类错误数据和误差数据引起的噪声问题,提出了一种基于小波阈值的数据去噪算法。通过构造小波重构系数并借鉴噪声方差和信噪比等指标,进行小波基函数筛选来确定适合于浮动车数据去噪的小波基函数,进而通过构造阈值函数来克服软硬阈值函数的不足,最后在实测数据上确定分解层数来实现一种针对浮动车数据的小波阈值去噪算法。通过实验,该算法在本文数据集上明显优于其他算法,能够较好的完成浮动车数据去噪。 考虑到目前道路交通事件检测均是以固定检测器数据为依托,使得检测结果的道路覆盖度较低且算法复杂较高的问题。本文基于去噪后的浮动车数据,通过对高速公路交通事件交通流特点进行分析,首先通过提取交通流参数构造正常数据排除算法,实现对待检测数据的第一层过滤来减少后续的计算量,进而借鉴小波理论在奇异性检测方面的优势,构造基于光滑小波函数的小波交通事件检测算法,实现对高速公路进行高精度、细粒度的交通事件检测。通过在实测数据上进行实验可知,本文方法的检测效果均优于其他算法,能够较好的完成道路交通事件检测。 在道路交通事件检测结果的基础上,通过对高速公路路网运行状态相关影响因素进行分析、提取,提出了一种基于径向基函数神经网络(Radial Basis Function Neural Network,RBFNN)的Bagging高速公路运行状态量化算法。Bagging通过对多个不同神经元个数的RBFNN进行集成,较好的实现了道路运行状态各因素与路网运行状态之间的复杂映射关系。通过以高速公路交通校调数据为依据,在实测数据上进行测试分析可知,该算法能够准确的进行量化,在本文数据上的量化效果明显优于其他算法。 关键词:浮动车;高速公路;交通事件检测;小波分析;道路交通运行状态量化

英文摘要

The rapid development of the transportation and the substantial increase in the demand for residents' travels have led to an increasing pressure on road traffic and transportation, especially on the inter-city highway and highway around the big cities. Once the traffic incident occurs, it will enormously affect the road patency even lead to a chain effect, which will have a huge impact on the highway transport capacity. Therefore, in order to realize the effective supervision of the status of highway, it is necessary to intensively study the key problems of highway network traffic state. Thus, based on the floating car trajectory data collected on the highway, this paper progressively studies the highway road information on three aspects: the floating vehicle data denoising, highway road traffic incident detection and highway traffic conditions quantification. The specific research contents are as follows: The temporal and spatial distribution of floating vehicle data determines the feasibility and scope of the research on the status of highway network. Although the massive trajectory data is collected, the strong randomness of the vehicle trajectory leads to uneven distribution of data in time and space, so it is difficult to define the scope of the study. In order to solve this problem, the range of road research and the cycle of road information release are determined by studying the lower limit of data required for road traffic information and the analysis of spatial and temporal distribution of data. In order to solve the problem of noise caused by various types of erroneous data and error data in the collection process of floating vehicle data, a data denoising algorithm based on wavelet threshold is proposed. Constructing the wavelet reconstruction factor, referring to the parameters such as noise variance and signal-to-noise ratio, wavelet basis function filtering is used 6to determine the wavelet basis function suitable for de-noising of floating vehicle data. Then a threshold function was constructed to overcome the shortcomings of soft and hard threshold function. Finally, the number of decomposed layers is determined on the measured data to realize the wavelet threshold denoising algorithm for the floating vehicle data. The experiment proved that the algorithm is superior to other algorithms in the data set of this paper, which can better complete the de-noising. Considering that the current traffic incident detection is based on fixed monitor data, which makes the test result of the road coverage is low and the algorithm is more complex. Based on the floating car data of de-noising, this paper analyzes the traffic flow characteristics of traffic incident in highway. Firstly, the normal data elimination algorithm is used to filter the traffic incident data to reduce the subsequent calculation. Then, based on the advantages of wavelet theory in singularity detection, the wavelet traffic incident detection algorithm based on smooth wavelet function is constructed to realize high precision and fine granular traffic incident detection on highway. The experimental results show that the proposed method is superior to other algorithms, and can better complete the road traffic incident detection. Based on the results of road traffic incident detection and through the analysis and extraction of the related factors related to the running state of the highway network, a new Radial Basis Function Neural Network-based bagging algorithm for quantization of highway state is proposed. Bagging integrates several RBF neural networks with different number of neurons, and achieves the relationship between the factors of road running state and its complicated mapping. The comparison of proposed algorithm's test result and the highway traffic calibration data shows that the algorithm can quantify accurately, and the quantization effect on the data is superior to other algorithms. Key Words: floating car data; highway; tyraffic incident detection; wavelet analysis; quantification of road traffic state

作者相关
主题相关
看过该书的人还在看哪些书