Abstract
Theoretically speaking, the data of a stated preference survey could be suggested for the calibration of a stochastic route choice model. However, it is unrealistic to implement the questionnaire survey for such a large number of alternative routes. Engineers generally determine the parameter empirically. This experienced choice of perception parameter may cause higher errors in the route flows. In our calibration model of the perception parameter, the data of the cellular network is set as the input. This model consists of two levels. The upper level is to minimize the gap squares of the route choice ratio between the C-logit model and the cellular network data. The stochastic user equilibrium (SUE) in terms of the C-logit model is used as the lower level. The simulated annealing (SA) algorithm is used to solve the model, where the route-based gradient projection (GP) algorithm is used to solve the inner SUE. A case study is used to validate the convergence of the model calibration. A real-world road network is used to demonstrate the objective advantage of an equilibrium constraint over a nonequilibrium constraint and explain the feasibility of the candidate routes assumption.
Highlights
The traffic conditions on a road network under continuous change from minute to minute, and it is hard for travelers to forecast the traffic in the future or understand the real-time traffic situation
The Root Mean Square Error (RMSE) is the difference between the ID path flow inferred in reverse from the traffic assignment model and the data records from the telecommunication company
The specific difference of the proposed model comes from the construction of the gap squared function, in which the flow of the base station antenna ID path was taken as the reference point and matched the covered road routes up to this cell phone trajectory
Summary
The traffic conditions on a road network under continuous change from minute to minute, and it is hard for travelers to forecast the traffic in the future or understand the real-time traffic situation. A stochastic route choice model requires a random variable representing the error term being determined, i.e., the difference between the perceived travel time and the actual one. MM algorithms are generally used to infer from GPS data based on the corresponding elements in the transportation network, including locations, links, and paths. The one-by-one route matching method is not suitable in this condition even though it is available in the route choice model with the GPS data. Unlike the previous logit route choice model of maximizing the likelihood function, in the proposed method, the gap of the route choice results and the cell phone trajectory data were minimized to obtain the parameter of the route choice model. Hand,On this the alternativework for the matching an ID path On andthe a road thework othercontributes hand, this to work perception calibration of a C-logit route choice.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
