01514303

Component

This research aims to make a comprehensive and objective comparison and assessment between the relative significance of three crucial issues in the context of probabilistic network analysis and prediction of static stochastic pattern of flow. In recent years, there has been a growing recognition towards the advantage of path-based methods of stochastic traffic assignment (STA). It has been well established that explicit treatment of path-flow variables allows application of more advanced random utility models in analysis of transportation networks. This research is intended to investigate the question that to what extent the size of generated path sets can affect prediction of flows. A simulated path generation algorithm was applied to an illustrative network and the investigations showed that the size of generated path sets only slightly affects the flow prediction. Further investigations have also been conducted with respect to the relative importance of model estimation and model specification in STA. In the literature of univariate STA, there has been considerable attention to utilization of sophisticated choice models to address different problems in this context including the problem of path overlapping. However, in most of these researches the problem of estimating the single calibration parameter of these models has been neglected and the models were studied using pre-specified parameters. A structured-parameter paired combinatorial logit (PCL) assignment model along with a practical and heuristic method of calibration is proposed in this study. The proposed model not only can represent the correlation among path utilities, but also considers the fact that traveling between different origin-destination (O-D) pairs may correspond to different levels of stochasticity. Results showed that calibration of stochastic user equilibrium (SUE) models could more meaningfully affect the prediction of flow than selection (specification) of the choice model.

This paper was sponsored by TRB committee ADB30(2) Network Equilibrium Modeling. Alternate title: Path-Based Stochastic Traffic Assignment: Investigation of the Effect of Choice-Set Size, Model Specification, and Model Calibration on Prediction of Static Flow.

01503729

14-0578

English

17p

2014

Transportation Research Board 93rd Annual Meeting

Digital/other

Figures; References; Tables

Calibration; Heuristic methods; Logits; Origin and destination; Stochastic processes; Traffic assignment; Traffic flow

Path based algorithms; Stochastic user equilibrium models

Highways; Planning and Forecasting; I71: Traffic Theory; I72: Traffic and Transport Planning

Transportation Research Board Annual Meeting 2014 Paper #14-0578

TRIS, TRB, ATRI

Jan 27 2014 2:16PM