A Modified Stochastic User Equilibrium Based Back-Propagation Method of Transportation Network State Estimation

In this research, we propose a modified stochastic user equilibrium based back-propagation method (MSUEBPM) to simultaneously estimate the transportation network state including an origin-destination (OD) demand matrix, link flow, and coefficients in the logit-model-based SUE. It is assumed that travelers in the same zone have similar route choice behavior. Multi-type data are required in this work, i.e. population, total demand of travelers, OD demand, link travel time and link flow. They can be derived from the residential trip survey, smartphone cellular signaling data, global position system, traffic sensors and so on. A back-propagation algorithm is applied to minimize a composite and non-convex objective function. The first-order partial derivatives of the objective function on the estimated coefficients are obtained. The proposed algorithm is tested in a simple network and a large network. the estimated results are good enough for engineering usage. However, in the large network, the calculation speed is much slower for the much more routes.