The Research On Primary Technologies In Real-time Traffic Information System Based On Big Data In Internet Of Vehicles

As the number of automobiles grows,traffic congestion is caused by the serious conflict that exists between vehicles and road.The appearance of Internet of Vehicles and Intelligent Traffic System can effectively solve the problems of traffic congestion that have seriously affected people’s life.Under this background,the relevant technologies of real-time traffic information system based on big data in Internet of Vehicles are studied in this thesis.The real-time and accurate traffic information can be acquired by processing and analyzing traffic data,and the information can help people choose reasonable travel routes and ease traffic jams.The main study contents of this thesis are presented as follows:Firstly,a carrier-cloud-client architecture based on Internet of Vehicles is studied to gain the realtime traffic information.The overall architecture of the system is firstly designed and then the hierarchical function descriptions of it are made in the sequence of traffic information collection,preprocessing,cloud platform and data mining.Based on the characteristics of the raw traffic data,some corresponding cleaning rules are made to deal with wrong,lost and redundant traffic data.Secondly,a dynamically adaptive algorithm of repairing data based on the temporal and spatial correlation of traffic data is proposed for repairing missed data lost in the process of collecting and cleaning data.In order to improve the quality of data repairing,the algorithm combines a secondary exponential smoothing method with the spatial correlation of traffic data.The simulation results show that the algorithm is of greater quality on data repairing.Finally,an improved fuzzy c-means algorithm used to identify traffic state is proposed to cluster massive traffic data.The Canopy algorithm and the Xie-Beni index are utilized to improve the algorithm.The Canopy algorithm is used to initialize the number of clusters and the cluster centers,and the Xie-Beni index is introduced to determine the fuzzy weighted index.The improved algorithm is used to cluster traffic data firstly,and then the algorithm of traffic pattern recognition and the criteria of traffic state classification are used to estimate traffic state.The simulation results show that the algorithm has smaller misclassification probability and can judge the traffic state well.