Study On Traffic State Identification And Prediction Of The Urban Periphery Road Network Of Intercity Freeway And Urban Expressway

With the changing of urban form and the space scale expansion, as well as with rapid increasing of the volume of motor vehicles, a number of urban freeway and expressway in periphery area of the metropolitan have been evolved into part of the road commuting of city. This situation resulted in low capacity and level of service in the peak hour, severe traffic congestion, frequent traffic accidents, etc. This phenomenon brought negative influence and constricted the freeway's function exertion such as smart, high efficiency and safety, and seriously hindered the development of inter-city transport. From the viewpoint of traffic operation, the urban periphery road network has become the new network traffic bottlenecks. Therefore, it is urgent to strengthen dynamic monitoring the traffic flow of these sections.In view of the techniques for analyzing traffic state, nowadays there is lacking of comparative mature comprehensive road traffic models and methods that composed of speed, occupancy and volume to determine the level of service of the automobile road, such as freeway and urban expressway. Therefore, it is difficult to reflect the real-time traffic conditions of the freeway which into or out of city. Existing freeway traffic monitoring systems establish strategy with few support from the accurate and reasonable theoretical basis or methodology. As a result, effective control strategy and means are not be taken in time, which often causes serious traffic congestion and obstruction.In this paper, Beijing freeway especially the the urban periphery road network of urban expressway and freeway is researched. Traffic characters are analysed based on the preprocessd data from the existing monitoring system. Then we use the integrated methods of fuzzy clustering and wavelet analysis to divide the state of urban expressway and freeway separately. And the domains of traffic flow parameters (volume, speed and occupancy) of different states (free flow, synchronized flow and wide moving jammed flow) are gained. At the same time, we establish state classification criterion of the urban periphery road network based on the traffic state of the freeway. Finally, we choose RBF neural network method to forecast volume, speed, and occupancy, then we predict the traffic state based on the forecasting results of traffic flow parameters. The prediction of the traffic state and the recognition result based on the actual data are compared. It can be concluded that the performance of the method for forecasting traffic situation based on traffic flow parameters is well, which has high congestion detection rate, low false alarm rate, short average detection time.