Abstract:

A study was performed to estimate the real-time vehicle queue length before the stop line at the urban signal controlled intersection. The relationship among traffic flow directions in road sections was analyzed, and two models,i.e., an artificial neural network model and a Bayesian network model were established to deal with the effects of traffic flows in the road sections, and the structure of the models was specified. To facilitate the practical application of these models, the principal component analysis was used to decrease the input dimensions, the EM algorithm and GMM were used to represent the parameters of the train models. A simulation road network was designed based on the real world data to validate the proposed models under various experiment scenarios. Simulation results showed that the Bayesian network model can grasp the traffic flow tendencies better than the artificial neural network model because of the randomness in the urban road network.

Key words: engineering of communications and transportation system, intersection, queue, artificial neural network, Bayesian network, principal component analysis