Theory And Method Studies On Traffic Situation Assessment For Urban Expressways

With the continued expansion of urban scale, expressways have become key factors that support the entire urban road traffic running, playing an irreplaceable role in urban transportation systems. Although the function of expressways is positioned to serve the long-distance and high-speed traffic, congestions caused by the increasing traffic demand inevitably happen on expressways. The relatively closed structure of the expressway network blocks the release of jammed traffic, resulting in more apparent effect of congestions in both temporal and spacialdimensions. Due to the important function and position of expressways, controlling and solving the congestion problems on the expressway network has become particularly urgent. Based on the available traffic data resources, the keys to the traffic managements on experessways are to quickly capture the external characteristics and evolutions of traffic, to accurately evaluate the traffic condition and potential capacity, and to reasonably predict the developing trend of congestions.In this context, this dissertation is intended to study key theories and technologies, by responding to practical problems and needs of the expressway network in Beijing, attempting to support traffic management and controls, with an emphasis on the macroscopic characteristics and dynamic evolution of the traffic stream on expressways,In order to completely understand the traffic characteristics on expressway networks, the traffic stream relationship model for expressways is calibrated based on the actual data; the time-varying correlation between traffic flow and occupancy and their periodic feature on different workdays are comparatively analyzed. Then, the recurrent severely congested segments on expressway networks during the peak hours are obtained using the method of time rate screening, and the temporal-spatial congested rate on typical expressway links is analyzed. The survival function and Lognormal distribution function are introduced to analyze the duration distribution characteristics of recurrent congestions, and the temporal-spatial speed distribution charts are used to visually show the evolution of traffic congestions.In order to realize the forecasting and early warning of congestions on expressways, the traffic estimation and prediction model of the traffic evolution on expressways is developed based on the modified cell transmission model. Based on the network structure and traffic characteristics of expressways, the links are divided into cells of different lengths by introducing the parameter of variable cell length. The time interval of the clock update is expanded properly by proposing the concept of "virtual transmission capacity." Van Aerde traffic stream model, instead of the trapezoid model, is used to describe the non-linear relationship among traffic variables. Algorithms for estimating variables of the congestion intensity, breadth, duration, and extent are designed based on the basic traffic flow parameters deriving the modified cell transmission model. The nonlinear traffic state-spacial prediction model is developed for expressway networks by setting the traffic density as the state variable and the speed as the observational variable. In order to solve the nonlinear traffic model, the extended Kalman Filter is introduced to realize the state estimation and multistep prediction.In regards to the evaluation problems of macroscopic traffic conditions on expressway networks, the evaluation model of "Macroscopic Traffic Condition Indices (MTCI)" is developed based on the mileage distribution of travel speeds on the network. The vector algorithm is proposed to obtain the macroscopic fundamental diagram model of networks, and the relationship models between average flow, average density, and average speed are built at different times, to represent the actual traffic service level. Based on the macroscopic fundamental diagram, the macroscopic traffic conditions on expressways are divided into five levels. The distribution model of travel speeds under different network conditions is built after numerous statistical analyses, the evaluation method of MTCI is developed based on the percentile thresholds, which is tested and verified using the western3rd ring-road expressway network.Finally, the western3rd ring-road expressway network is selected to implement the traffic state multistep prediction. And then, the whole performance of the network is evaluated using the proposed evaluation model.From the practical traffic problems on the expressway network in Beijing, this dissertation completes four steps of theory analysis, feature study, model development, and model application. It also presents some recommendations on improving the model accuracy, expanding model applications, and others for further study, in order to support and serve the urban traffic managements using the reliable traffic situation assessment results.