Future Speed Prediction Of Vehicles

Speed prediction of vehicles has become one of the research highlight in the field of smart vehicles and intelligent transportation system. Speed prediction of vehicles based on Internet of Vehicles can be used for path navigation, vehicle collision warning and Eco system. So the research of speed prediction has an important theoretical and application value. The major contents of this paper are as follows:① Urban road network model was established by VISSIM software to simulate different traffic conditions, including congested, smooth and slow-speed condition. Thus, the original data for speed prediction were obtained. Path real-time division dynamic model was established, variable-length division and fixed-length division were combined to realize future road division of vehicles. Division method of basic sections and its sub-section synthesis were determined.② Two kind of gray speed prediction model were proposed, including simple model GM(1,1) and gray neural network model. The study results have shown that simple GM(1,1) can be applied to predict speed of the vehicle which tend to steady-speed driving, acceleration driving and deceleration driving, while gray neural network model can be used to predict speed when a vehicle driving under stop-and-go congested condition. The effects of different intervals on two kinds of gray prediction models were analyzed. Prediction precision of each gray speed model was verified.③ Most similar neighbor speed prediction model was established. Historical database was built by historical data of road average speed extracted on the basis of path division. Then the historical series of road average speed that most similar to the current road average speed variation was used as future average speed of road. The effects of different intervals and sample time on most similar neighbor speed prediction model were analyzed, and its prediction precision was verified.④ Based on the data of speed, acceleration and vehicle gap distribution of all vehicles on four typical conditions, box plots of seven basic characteristic parameters of four different typical conditions were analyzed. Road average speed and speed variance ware chosen for condition recognition and recognition judgment rules were made base on the two parameters.Finally, the applicable conditions and scopes of four speed prediction methods were proposed, including the model of gray speed prediction, the model of most similar neighbor speed prediction, the approximate prediction method by speed track of other vehicles, the approximate prediction method by road average speed...