Study on Micro-behaviour of Interference between Bicycle and Motor-Vehicle at Signalized Intersection Based on BP Neual Network

The crossing behavior of motor vehicles and bicycles at signalized intersections is of great importance in the mixed traffic microscopic simulation. However, it is very difficult to simulate the behavior because of the great variations in the driver behaviors, vehicle characteristics and traffic environment. The current study proposes a neural network (NN) approach to simulate the right-turning motor vehicles passing through straight-moving bicycles. A computer based three-layered back-propagation neural networks (BPN) model was developed for the estimation of motor vehiclespsila crossing decision mode. The BPN model was trained, validated with field data and then compared with the discrete choice logistic model. It was found that the BPN model performed better. Result showed that the proposed model could produce reasonable crossing decision mode estimates for individual vehicles.