Simulation and modeling traffic flow based on Division K Nearest Neighbor

Abstract

A new traffic flow model is proposed based on cellular automata and Division K Nearest Neighbor for the predication problem of traffic flow state change trend. The model firstly gives the update rules of vehicle state evolution and lane change rules of a vehicle, and establishes the state prediction model based on Division K Nearest Neighbor. Finally, the simulation analysis is conducted by the use of experimental platform, and the relationship between the factors such as average traffic flow speed, average flow rate, traffic flow density and lane change frequency, etc. is deeply studied. The results show that the prediction model has great advantages in the medium and low density area, and lower lane change rate has limited effect on the improvement of traffic flow.