本文選題：城市道路交通運行狀態(tài) + 時空相關(guān)性　； 參考：《北京交通大學》2017年博士論文

【摘要】：城市道路交通流特征分析、交通運行狀態(tài)綜合評價和預測理論及其應用研究是智能交通系統(tǒng)建設和發(fā)展的基礎(chǔ)和關(guān)鍵,不僅有助于交通管理部門全面掌握路網(wǎng)的整體運行情況、分析交通擁堵成因及其變化規(guī)律,而且有助于出行者及時了解交通運行信息,以便躲避交通擁堵路段,降低出行成本。這些對于緩解城市交通擁堵、提升交通管理效率和交通運行信息服務質(zhì)量具有非常重要的理論意義和應用價值。本文圍繞緩解城市道路交通擁堵問題,面向交通管理和公眾出行的實際需求,結(jié)合采集的交通流基礎(chǔ)數(shù)據(jù),對道路交通流時空特性分析、路網(wǎng)關(guān)鍵路段識別、路網(wǎng)交通運行狀態(tài)綜合評價、交通運行狀態(tài)長時預測及其理論成果的實際應用等問題進行了深入的研究,形成了系統(tǒng)的道路交通運行狀態(tài)分析的體系方法架構(gòu)。最后通過實際應用案例分析,驗證了研究成果的可行性和實用性。本論文的主要研究成果包括以下幾個方面:(1)以道路交通流的時空特性為基礎(chǔ),提出了基于時空相關(guān)性的道路交通流時空特征分析方法,實現(xiàn)了對路網(wǎng)關(guān)鍵路段的有效識別。以時空相關(guān)函數(shù)作為量化指標,從時間和空間兩個角度對路段交通運行狀態(tài)之間的相互影響關(guān)系進行了分析。該方法首先根據(jù)復雜網(wǎng)絡理論和交通流的影響傳播關(guān)系,建立了路網(wǎng)空間鄰接矩陣;然后引入空間延遲算子,從整體和局部兩個層次定義了時空相關(guān)函數(shù)的計算方法,并分別對整體路網(wǎng)和局部路段在不同時間延遲和空間延遲下交通運行狀態(tài)之間相互影響的動態(tài)變化規(guī)律進行了分析;最后將路段的重要性定義為路段與周邊相鄰路段交通狀態(tài)之間相互影響程度的大小,利用逼近于理想點的排序方法,通過局部時空相關(guān)函數(shù)計算每個路段與理想點之間的貼近程度,最終實現(xiàn)對路段重要性的度量。實驗結(jié)果表明該方法計算簡單,能夠?qū)β肪W(wǎng)關(guān)鍵路段進行有效識別,具有可行性和實用性,為構(gòu)建交通運行狀態(tài)綜合評價和預測模型提供了重要的理論依據(jù)。(2)以獲取的實際交通流數(shù)據(jù)為基礎(chǔ),構(gòu)建了滿足不同需求的多層次的城市道路交通運行狀態(tài)評價指標體系。針對交通管理、交通運行狀態(tài)評價以及出行信息服務的多層次需求,確定了評價指標的選取原則和體系結(jié)構(gòu),構(gòu)建了城市道路交通運行狀態(tài)評價指標體系,從微觀路段、中觀道路、宏觀路網(wǎng)三個層次對交通運行狀態(tài)在時間和空間上的變化特征進行描述。然后根據(jù)路網(wǎng)結(jié)構(gòu)特征,分別確定相應的微觀路段、中觀道路和宏觀路網(wǎng)的評價指標,并對各個評價指標的定義、取值以及相關(guān)計算方法進行了說明。(3)以城市道路交通運行狀態(tài)評價指標體系為基礎(chǔ),提出了基于模糊熵-熵權(quán)法的交通運行狀態(tài)多指標綜合評價方法。該方法充分考慮交通流的復雜性和模糊性,運用FCM聚類分析方法將交通運行狀態(tài)劃分為非常暢通、基本暢通、穩(wěn)定通行、緩慢通行、一般擁堵和非常擁堵6個不同的狀態(tài)等級,并確定了微觀、中觀指標對應不同狀態(tài)等級的模糊區(qū)間范圍。然后利用直覺模糊熵理論,將模糊區(qū)間轉(zhuǎn)化為直覺模糊數(shù),以此作為評價模型的輸入,選取路段重要性和流量公里數(shù)作為評價模型的權(quán)重。綜合運用多個評價指標,從微觀路段、中觀道路、宏觀路網(wǎng)三個層次對交通運行狀態(tài)進行了量化,實現(xiàn)了定性分級描述和定量分析的有機結(jié)合。實例分析表明評價結(jié)果符合人們的認知,能夠滿足交通管理和公眾出行的需求。(4)以道路交通流時空相關(guān)性的周期性變化特征為基礎(chǔ),構(gòu)建了基于非參數(shù)回歸核估計的交通運行狀態(tài)長時預測模型。該方法以預測交通運行狀態(tài)綜合評價結(jié)果(綜合評價指數(shù))為目標,采用非參數(shù)回歸的方法來估計未來時刻交通運行狀態(tài)的分布。首先根據(jù)預測目標與周邊路段交通流之間時空相關(guān)性的大小,從時間、空間兩個角度分析了路網(wǎng)其他路段對預測目標的影響,確立了模型輸入的時空變量;然后利用函數(shù)型主成分分析的貼近度度量方法,對不同周期交通流時間序列之間的長期變化趨勢的相似性進行了計算;最后通過選取核函數(shù)和最優(yōu)窗寬,實現(xiàn)了對交通運行狀態(tài)的長時預測。實驗結(jié)果表明,該模型具有較高的預測精度,適用于其他不同層次評價指標的預測,具有普適性和可擴展性。(5)將論文理論研究與實際應用相結(jié)合,提出了北京市道路交通運行信息服務的應用案例分析。立足北京市交通管理部門和出行者的實際需求,提出了面向交通管理的交通運行信息評價統(tǒng)計服務和面向公眾出行的交通運行信息預測預報服務,并對信息發(fā)布服務的應用方案進行了設計。以面向北京市的實際應用案例,驗證了論文理論研究成果的實用性和有效性。
[Abstract]:The characteristic analysis of urban road traffic flow, the comprehensive evaluation and prediction theory of traffic operation state and its application are the foundation and key of the construction and development of the intelligent transportation system. It not only helps the traffic management department to master the overall operation of the road network in an all-round way, analyzes the cause of traffic congestion and the law of change, but also helps the travelers to be in time. In order to avoid traffic congestion and reduce travel costs, it is of great theoretical significance and application value to alleviate traffic congestion, improve traffic management efficiency and traffic information service quality. This paper focuses on alleviating traffic congestion in urban roads, facing traffic management and public exit. The actual demand of the line is combined with the basic data of the traffic flow, the analysis of the spatial and temporal characteristics of the road traffic flow, the identification of the key sections of the road network, the comprehensive evaluation of the traffic operation state of the road network, the long time forecast of the traffic operation and the practical application of the theoretical results. In the end, the main research results of this paper include the following aspects: (1) based on the spatio-temporal characteristics of road traffic flow, a spatio-temporal feature analysis method based on spatio-temporal correlation is proposed to achieve the opposite path. The effective identification of key sections of the network is taken as the quantitative index of spatio-temporal correlation function. The interaction relationship between the traffic state of the road is analyzed from two aspects of time and space. First, the network space adjacency matrix is established based on the complex network theory and the influence of traffic flow, and then the space extension is introduced. Late operator, the calculation method of spatio-temporal correlation function is defined from the two levels of the whole and the local, and the dynamic change law of the interaction between the whole road network and the local section under the different time delay and the space delay is analyzed. Finally, the importance of the section is defined as the link and adjacent sections. The degree of mutual influence between traffic states is used to calculate the close degree between each section and the ideal point by the local spatio-temporal correlation function. The results of the experiment show that the method is simple and can effectively identify the key sections of the road network. It is feasible and practical, and provides an important theoretical basis for the construction of comprehensive evaluation and prediction model of traffic operation state. (2) based on the actual traffic flow data obtained, a multi-level urban road traffic operation state evaluation index system is constructed to meet the different needs. According to the multi-level demand of information service, the selection principle and system structure of evaluation index are determined. The evaluation index system of urban road traffic operation state is constructed. The change characteristics of traffic operation state in time and space are described from three levels of micro section, meso road and macro road network. Then, according to the characteristics of road network structure, The evaluation indexes of the corresponding micro sections, the meso road and the macro road network are determined respectively, and the definitions, values and relevant calculation methods of each evaluation index are explained. (3) based on the evaluation index system of urban road traffic operation state, a multi index comprehensive evaluation of traffic operation state based on fuzzy entropy entropy weight method is proposed. This method takes full consideration of the complexity and fuzziness of traffic flow, and uses the FCM clustering analysis method to divide the traffic state into 6 different state levels, namely, very smooth, smooth, stable, slow, common congestion and very congested, and determines the fuzzy interval model of the microscopic, meso index corresponding to different state levels. Then, using the intuitionistic fuzzy entropy theory, the fuzzy interval is transformed into the intuitionistic fuzzy number. As the input of the evaluation model, the weight of the link importance and the number of traffic kilometers is selected as the weight of the evaluation model. The traffic operation state is quantified from three levels, the micro section, the meso road and the macro view network. It realizes the organic combination of qualitative classification description and quantitative analysis. The case analysis shows that the evaluation results conform to people's cognition and can meet the needs of traffic management and public travel. (4) based on the periodic variation characteristics of the spatio-temporal correlation of road traffic flow, a long time preview of traffic operation state based on non parametric regression kernel estimation is constructed. The method uses the non parametric regression method to estimate the distribution of traffic running state in the future by using the method of non parametric regression to predict the distribution of traffic state in the future. Firstly, the road network is analyzed from two angles of time and space, according to the big small spatio-temporal correlation between the predicted target and the traffic flow of the surrounding sections. The spatial and temporal variables of the model are established by the impact of the section on the prediction target, and then the similarity between the time series of different cycle traffic flows is calculated by using the proximity measurement method of the functional principal component analysis. Finally, the traffic operation state is realized by selecting the number of kernel functions and the optimal window width. Long time prediction. The experimental results show that the model has high prediction accuracy and is suitable for other different levels of evaluation index prediction, it is universal and extensible. (5) combining the theoretical research and practical application of the paper, the application case analysis of Beijing city road traffic operation information service is put forward. Based on the Beijing traffic management department The actual demand of the door and the traveler, put forward the traffic operation information evaluation and statistics service oriented to traffic management and the traffic operation information forecast service oriented to the public travel, and design the application scheme of the information publishing service. The practical application case of Beijing is used to verify the practicability of the theoretical research results of the paper. And effectiveness.

【學位授予單位】：北京交通大學
【學位級別】：博士
【學位授予年份】：2017
【分類號】：U491
，

本文編號：1813627