This study presents a data-driven approach to addressing traffic congestion on toll roads by proposing a dynamic traffic allocation method designed to optimize bottleneck management. Leveraging both sensor and historical traffic data, a Graph Neural Network (GNN)-based model is developed to monitor key metrics such as traffic volume and speed in real time, enabling precise identification of congestion points along toll roads. In conjunction with this, a dynamic traffic assignment model is formulated, incorporating Lagrangian Relaxation with User Equilibrium (UE) and System Optimal (SO) theory, to dynamically adjust traffic distribution and alleviate pressure on congested segments. The proposed framework is validated on a toll highway using peak-hour traffic data. Experimental results demonstrate that the model effectively enhances toll road performance, with average speeds on bottleneck sections increasing by 15-25% and the congestion index decreasing by approximately 30%. Furthermore, by optimizing toll strategies and rerouting traffic, the dynamic assignment model significantly reduces queue times during peak traffic periods. Integration of Graph Neural Networks with real-time toll road management. The contribution of this study lies in its utilizing multi-source data for accurate congestion prediction and adaptive traffic flow optimization. This approach not only offers theoretical advancements but also provides a practical tool for improving toll road operations, making it highly valuable for realworld applications.
Bottleneck identification and dynamic traffic assignment on toll roads
International Conference on Smart Transportation and City Engineering (STCE 2024) ; 2024 ; Chongqing, China
Proc. SPIE ; 13575
2025-04-28
Conference paper
Electronic Resource
English
Dynamic Toll Pricing using Dynamic Traffic Assignment System with Online Calibration
Transportation Research Record | 2019
|Engineering Index Backfile | 1931
|Engineering Index Backfile | 1928
|