Research On Short-term Traffic Flow Prediction Problems

Urban traffic problems have become one of the world's problems.Intelligent transportation system can better solve this problem.Intelligent transportation system has many advantages,both to ease traffic congestion,but also to improve the efficiency of passage.And it can reduce environmental pollution and so on.Short-term traffic flow forecasting is one of the important functions of intelligent transportation system.The traditional short-term traffic flow forecast is often influenced by the small data acquisition and weak computer performance.With the development of science and technology,the increasing of traffic data acquisition technology and transmission equipment,we obtain data from many ways,and we can get a lot of real-time data.Moreover,in recent years,with the development of the cluster,the computing power of the computer is greatly improved,and it can deal with the large amount of data and the complicated model,and can realize the analysis and prediction of the short-term traffic flow data.The main contents of this paper are:(1)Due to the failure of the detector and other reasons,data may be missing and redundant,this paper detects and repair the anomaly and missing data.(2)Considering the periodicity of traffic data,the ARIMA model is used to predict the short-term traffic flow.(3)Considering the impact of traffic flow on the upstream and downstream sections,the network topology characteristics are used by SVM model.Compared with historical information of the current intersection,there is a better performance.(4)We divide 24 hours into 4 linear stages according to the bimodal distribution of traffic flow.And we build one model for everyone.