Real-time Vehicle Velocity Prediction Strategy under Highway Vehicle-to-vehicle Environment

Abstract

To improve the prediction accuracy and calculation speed of vehicle velocity prediction under a highway vehicle-to-vehicle (V2V) environment, a velocity prediction strategy combined with traffic information is proposed under the corresponding scenario. First, the highway scenario is set up. In addition, the general regressive neural network (GRNN) combined with front vehicle information is used in the highway scenario, which constructs a GRNN velocity prediction model with multi-information fusion. Then the model in the highway scenario is simulated, and the optimal model parameters are extracted, which are used to verify the prediction model in other operating conditions. As a result, the simulation data shows that compared with the prediction strategy using only historical vehicle velocity data, the prediction accuracy of the fusion prediction strategy using the optimal parameters is improved by 14.4% in the highway scenario under the V2V environment.