Real time traffic signal learning control using BPNN based on prediction for probabilistic distribution of standing vehicles

In this paper, a new method to predict the probabilistic distribution of traffic jam at crossroads and a traffic signal learning control system are proposed. First, the Dynamic Bayesian Network is used for build a forecasting model to predict the probabilistic distribution of vehicles for traffic jam during the each period of traffic signal. The adjusting algorithm of traffic signal control is applied to maintain the probability of a lower limit and ceiling of the standing vehicles to get the desired probabilistic distribution of the standing vehicles. In order to achieve the real time control, a learning control system based on the Back Propagation Neural Network is used. Finally, the effectiveness of the new traffic signal control system using the actual traffic data will be shown.