X.-X. Weng and G.-L. Du (PRC)
Modeling, neural network, hybrid Elman, traffic prediction, short-term traffic flow
Accurate and timely prediction of traffic flow is paramount for effective traffic signal control and traveler information dissemination. A novel prediction model based on a hybrid Elman neural network for short-term traffic flow of urban intersections is presented. By analyzing characteristics of short-term traffic system with strong uncertainness and non-linearity, the period of signal control is taken as time interval of model and flow rate and occupancy within green phase are taken as 2 dimension variables of the model. Information of future and past prediction system is fused. The model that depends on the feedback of connective layer strengthens memory of network for time series feature’s information. The experimental research indicates that predictive accuracy of the model is more effective and advanced than traditional methods.
Important Links:
Go Back
