Research on Level of Service Prediction for Expressway Work Zones

2026-99-2417

11/20/2026

Authors
Abstract
Content
In expressway work zones, lane closures drastically lower the level of service (LOS) of road segments, resulting in transient bottlenecks that jeopardize traffic safety and network efficiency. A workable technique that can quickly and effectively forecast LOS in dynamic work-zone contexts is desperately needed to support scientific and effective traffic-diversion planning. This paper suggests an integrated prediction system that combines ensemble learning with microscopic traffic simulation. First, the VISSIM platform was used to create a high-fidelity simulation experiment matrix. The intended speed distribution was calibrated using field data to make sure the model accurately depicts the features of a genuine highway. The simulated speed distribution was similar with observed traffic, according to the findings of the Kolmogorov–Smirnov (K–S) test (D=0.038, p>0.05) and a Pearson correlation coefficient of 0.70.Five major influencing factors—heavy vehicle fraction (10%–40%), pre-construction traffic volume (25,000–100,000 pcu/d), number of lanes (2–5), speed restrictions (60–120 km/h), and work-zone lengths (1,000–4,000 m)—were combined to create 1,024 distinct work-zone scenarios. In order to capture intricate nonlinear coupling relationships, a Stacking ensemble prediction model was then created with XGBoost as the main learner with Support Vector Machine (SVM) and Random Forest (RF) as supplements. The suggested ensemble model performs noticeably better than single-algorithm approaches, according to experimental results. The model significantly outperformed the baseline XGBoost model’s accuracy of 72.75% on the test set, with an overall accuracy of 88.57%, a recall of 88.09%, and an F1-score of 87.85%.The model’s excellent generalization ability and dependability as a quantitative instrument for comparing traffic organization schemes and assessing the efficacy of pre-construction diversion methods are demonstrated via case validation.
Meta TagsDetails
DOI
https://doi.org/10.4271/2026-99-2417
Citation
Wang, L., Jiang, R., Zhang, G., Li, Y., et al., "Research on Level of Service Prediction for Expressway Work Zones," 2026 2nd International Conference on Intelligent Transportation and Future Mobility (ITFM2026), Xi'An, China, March 27, 2026, https://doi.org/10.4271/2026-99-2417.
Additional Details
Publisher
Published
Yesterday
Product Code
2026-99-2417
Content Type
Technical Paper
Language
English