A diffusion model to explain and forecast freeway breakdown and delay

摘要

The chance that a freeway will breakdown, transition from a free-flow to a congested state, is normally assumed to increase with an increase in traffic volume V (vehicles per unit time). In this paper, this assumption is challenged. Traffic density K (vehicles per unit length) proves to be a better predictor. Diffusion or stochastic differential equation (SDE) modeling is used to substantiate the claim. SDE modeling is especially useful in explaining the role that traffic noise (volatility) plays in breakdown. The SDE models take advantage of the unique properties of the geometric Brownian motion (gBM) and Ornstein-Uhlenbeck (OU) model structures. The breakdown probability model of π (K) and delay models provide accurate forecasts.