Study On Forecast Method Of Freestyle Road Network Traffic Congestion

With the continuous development of economy and the increasing urbanization, car ownership increased year by year, The problems of modern urban transportation are more and more serious, such as traffic congestion, traffic pollution and traffic accidents, etc.. These problems seriously affect people’s travel, reduce the efficiency of travel, urban road traffic is given more and more pressure. In response to the urban traffic problems, dynamic traffic management based on the intelligent transportation system(ITS) gets rapid development.Due to terrain, the phenomenon of cities with freestyle network is particularly complicated, the congestion is serious and difficult to evacuate. To ease the freestyle road network traffic congestion, It is necessary to research traffic flow short-term prediction, then according to the forecast results to determine traffic congestion identification and give a real-time traffic congestion forecast information.This paper mainly studies freestyle road network, introduces the basic characteristics of traffic flow parameters, analysis of traffic flow parameters acquisition technology and its scope of application, research the traffic information and processing methods. Research the traffic flow characteristics and space-time distribution characteristics of freestyle road network traffic congestion, in-depth analysis the cause of freestyle road network traffic congestion.For freestyle road network, If the way of road sections is far, the traffic flow parameters correlation is low, If the way of road sections is near, but due to traffic control measures(such as one-way traffic ban left or intersection),the traffic flow parameters correlation is also low. Accordingly, this paper research short-term traffic parameters prediction of freestyle road network based on the spatial correlation. Through the analysis of single forecasting method, this paper use multidimensional ARIMA prediction method based on the new rate algorithm and RBF neural network prediction method to predict freestyle road network traffic parameters for short-term. And combine these two kinds of forecast method based on the optimal power allocation, use combination model to forecast traffic parameters for short-term. Then, demonstrate the prediction model’s effect by example.According to the space-time distribution characteristics of traffic flow, This paper divide the freestyle road network traffic congestion into four grades, unblocked, slight congestion,congestion and jams. Respectively from Intersections and sections to select evaluation index,and constitute an index system. Use fuzzy comprehensive evaluation model based on AHP- Entropy to determine the weight of index to evaluate the prediction of traffic flow parameters, identification of traffic congestion, and issue traffic congestion forecast information in real time.The traffic forecast method of this paper researched can forecast the occurrence of traffic congestion in time. It has a certain reference value for traffic participants to choice transportation route and traffic management to make traffic evacuation decision.