Research On Traffic Prediction And Early Warning System Based On Data Mining

As people living standard enhancement and the development of the auto industry,the information of the degree of traffic flow and traffic safety has become the focus of people's daily life at any time.Predicting the traffic congestion and early warning the risk of traffic accurately can help people improve the quality of travel effectively.In this information era of "data explosion",using the various data obtained from the vehicle in the process of acquisition can get a lot of valuable information.In our country,the data mining technology in the application of intelligent transportation network is still in the preliminary stage,and traffic information will become one of the focus of the information society construction.Based on actual demand above,this paper discusses the two main technology in data mining,clustering analysis data mining and association rule data mining,covers each branch of both data mining algorithm,and briefly describes the definition of various algorithms,the basic thought and main mining steps.On the basis of the classical algorithms,this paper proposes an improved k-means clustering analysis algorithm and an improved Apriori association rules algorithm,illustrates the methods and steps of the improved algorithms,and compare the improved algorithms with the classic algorithms and the improved algorithms which mentioned in other papers in mining time and accuracy,highlight the advantage of the improved algorithms.As a data mining application scenarios,on the basis of the original vehicle network,this paper designs a traffic prediction and early warning system.The system consists of vehicle terminal,communication gateway(including database)and web platform.The various of data collected by car terminals go through the communication gateway to upload to the database.Then after further data filtering,data cleaning,data extraction and reloading,the data is processed by the improved data mining algorithms to mine the useful knowledge of traffic prediction and early warning.Then the knowledge is put back into the database.The web platform is a demonstration site of the traffic prediction and early warning system which based on J2EE development.In addition to basic information management functions,the main function of the web platform is to show the results of data mining on the GIS map,and to issue to the car terminals through the communication gateway.The traffic prediction and early warning system can predict the traffic congestion and warn the road risk effectively.To a certain extent,it provides a scientific support for traffic management.