Real time turning flow estimation based on model predictive control

In order to predict the real time turning flow at intersections, which is used for the real-time adaptive traffic signal control, a real time turning flow estimation model based on model predictive control is proposed. The model adopts multiple independent parallel BP neural networks to structure the prediction model in the model predictive control mechanism, which adequately exerts the advantages of rolling optimization, feedback correction, and multi-step prediction. The benefit of this is to improve the prediction accuracy. We utilize the microscopic traffic simulator with mathematical software and proper computational applications for the simulation. The simulation results prove that real time turning flow estimation model based on model predictive control ha s been more effective, compared with the traditional neural network prediction model.