Intelligent simulation and prediction of traffic flow dispersion

Dispersion of traffic flow on urban road segments is often described by some typical statistical models such as the normal distribution model and the geometric distribution model. These probability-based models can fit traffic flow well under ideal physical environments but may not work satisfactory in certain complex cases because of their strict mathematical assumptions. A neural network-based system identification approach is used to establish an auto-adaptive model for simulating traffic flow dispersion. This model, being feasible to a wide variety of traffic circumstances, can be calibrated and used for on-line traffic flow forecasting. Data simulation and field-testing show reliable performance of the proposed intelligent approach.