Abstract: The modeling and simulation of bicycle accident avoidance with other vehicles (motorcar/other bicycles/pedestrians) at non-signalized intersections is of great importance. However, it is very difficult to simulate the accident avoidance of individual bicycle because of the great variation, characteristics and traffic environment in cycling. The current study proposes a back propagation neural network (BPNN) approach to simulate the bicycle accident avoidance at nonsignalized intersections. A computer-based four-layered (NN) model was developed for bicycle accident avoidance modeling. The NN model was trained, validated with field data and then compared with other models. It was found that the NN model performed well. Results showed that the NN model could produce reasonable estimates for individual bicycle at non-signalized intersections.

Key words: bicycle; accident avoidance; neural network; nonsignalized intersections