Calibration of Dynamic Traffic Assignment Models with Point-to-Point Traffic Surveillance

Vaze, Vikrant; Antoniou, Constantinos; Yang, Wen; Moshe Ben‐Akiva · 2009 · OpenAlex-citations

DOI: 10.3141/2090-01

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This thesis addresses the critical need for accurate calibration of Dynamic Traffic Assignment (DTA) models to support efficient traffic management and traveler information systems. As highway infrastructure growth fails to keep pace with increasing traffic demand, optimizing existing networks through Intelligent Transportation Systems (ITS) has become essential. The research specifically investigates the calibration of both demand and supply parameters within DTA systems using data from emerging Automatic Vehicle Identification (AVI) technologies, which provide rich, disaggregate point-to-point traffic surveillance data. The study aims to demonstrate that incorporating AVI data significantly improves calibration accuracy compared to relying solely on conventional point sensor data. The methodology formulates the calibration problem within two distinct frameworks: a stochastic optimization framework and a state-space framework. To solve these formulations, the author implements and compares three specific algorithms: Simultaneous Perturbation Stochastic Approximation (SPSA), a gradient approximation-based path search method; a Genetic Algorithm (GA), a random search meta-heuristic; and a Particle Filter, a Monte-Carlo simulation-based technique. The research utilizes DynaMIT, a mesoscopic DTA system, as the model to be calibrated. Validation and testing are conducted in two stages. First, a small synthetic network is used to illustrate effectiveness, with synthetic sensor data generated via the microscopic simulator MITSIMLab. Second, the methodology is applied to a real-world case study of the Lower Westchester County (LWC) network in New York to demonstrate scalability. In this case study, MITSIMLab serves as a proxy for reality, generating necessary AVI surveillance data since actual AVI data streams were not yet available from the New York State Department of Transportation. The results indicate that the utilization of AVI data significantly enhances the accuracy of DTA model calibration. In both the synthetic network tests and the LWC case study, calibration procedures incorporating AVI data outperformed those using only conventional point sensor data. The study evaluated both demand-only calibration and simultaneous demand-supply calibration. Across all tested scenarios, the inclusion of point-to-point travel time measurements allowed for a more precise estimation of network states and parameters. The comparative assessment of the three algorithms provided insights into their respective performances within the defined frameworks, confirming the viability of the proposed methodologies for handling complex demand-supply interactions. The significance of this work lies in establishing a robust methodology for integrating emerging disaggregate traffic data into DTA systems. By demonstrating that AVI data improves calibration accuracy, the thesis supports the development of more reliable traffic prediction and management tools. This advancement is crucial for the effective operation of Advanced Traffic Management Systems (ATMS) and Advanced Traveler Information Systems (ATIS), which rely on accurate real-time network states to optimize traffic flow and guide traveler decisions. The study provides a foundation for future research into real-time calibration and the integration of diverse sensor technologies in transportation modeling.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• exposure measurement