Transportmetrica B: Transport Dynamics

Machine Learning for Macroscopic Traffic Modeling and Optimization

Introduction

Macroscopic traffic models are powerful yet efficient tools for analyzing traffic dynamics at large spatial and temporal scales. By abstracting from individual vehicle interactions, macroscopic models offer a computationally efficient and analytically versatile framework for traffic analysis at the network level. The balance between simplicity and explanatory power makes macroscopic traffic models particularly valuable for large-scale traffic analysis, especially in contexts where microscopic details are unnecessary or impractical. Despite the advantages, the reliability of macroscopic models has been constrained by the traditional statistical methods used for calibration, often limiting their accuracy and adaptability in complex real-world contexts. Recent advances in machine learning and data analytics offer new opportunities to address these challenges. The integration of data-driven techniques with macroscopic modelling offers substantial potential to improve predictive accuracy, enhance model robustness, and broaden practical applications. Enhancing macroscopic traffic models with machine learning techniques will further broaden their applications, ranging from transportation planning and infrastructure design to real-time traffic management and control.

Scope

This special issue aims to gather a diverse collection of contributions that advance the theoretical underpinnings, methodological innovations, and applied implementations of macroscopic traffic modeling enhanced by data-driven techniques. Emphasis will be placed on work that demonstrates how machine learning can complement, augment, and/or transform conventional models to address real-world transportation challenges, including traffic prediction, optimization, management, and control. Relevant topics include, but are not limited to:

• Generalizable macroscopic traffic modeling frameworks
• Network partitioning, classification, and clustering for scalable modeling
• Probe-based data fusion and state estimation
• Self-calibrating macroscopic simulation
• Explainable AI for macroscopic traffic modelling
• Uncertainty quantification in macroscopic traffic modelling
• Integration of machine learning with simulation platforms
• Data assimilation and imputation methods for large-scale traffic analysis
• Macroscopic impact analysis of connected and automated vehicles (CAVs)
• Centralized and decentralized cooperative control strategies for CAVs
• Traffic optimization and control leveraging connected vehicle data
• Dynamic congestion pricing enabled by predictive macroscopic model
• Real-time adaptive lane management strategies
• Real-time traffic prediction and control
• Staggered work-hour scheduling and demand management for congestion mitigation
• Parking design and pricing policies as macroscopic demand management tools
• Integration of macroscopic traffic models with energy and emission analysis

Keywords:

Machine Learning, Macroscopic Traffic Modeling, Reinforcement Learning, Traffic Management and Control, Connected and Automated Vehicles, Energy and Emission Analysis, Macrosimulation