A Collapsed Generalized Aw-Rascle-Zhang Model and Its Model Accuracy

TL;DR

This paper introduces the collapsed generalized Aw-Rascle-Zhang (CGARZ) model, a second-order traffic flow model that improves accuracy by capturing free-flow dynamics and congestion spread, validated against real-world data.

Contribution

The paper develops a new CGARZ model within the GSOM framework and provides a systematic calibration and validation approach using diverse traffic data.

Findings

CGARZ model offers a good balance of simplicity and accuracy.

It captures free-flow and congested traffic behaviors effectively.

Model validation shows competitive prediction accuracy.

Abstract

This work presents a collapsed generalized Aw-Rascle-Zhang (CGARZ) model, which fits into a generic second order model (GSOM) framework. GSOMs augment the evolution of the traffic density by a second state variable characterizing a property of vehicles or drivers. A cell transmission model for the numerical solution of GSOMs is derived, which is based on analyzing the sending and receiving functions of the traffic density and total property. The predictive accuracy of the CGARZ model is then compared to the classical first-order LWR and four second-order models. To that end, a systematic approach to calibrate model parameters from sensor flow-density data is introduced and applied to all models studied. The comparative model validation is conducted using two types of field data: vehicle trajectory data, and loop detector data. It is shown that the CGARZ model provides an intriguing combination of simple model dynamics in the free-flow regime and a representation of the spread of flow-density data in the congested regime, while possessing a competitive prediction accuracy.