Random Forests-Based Operational Status Perception Model in Extra-Long Highway Tunnels with Longitudinal Ventilation: A Case Study in China

Abstract

An increasing number of extra-long highway tunnels have been built and put into operation around the world, but the quantified segmentation criteria for evaluating the in-tunnel operational status have not yet been enacted up till the present moment. Meanwhile, ventilation facilities could not satisfy the dynamic requirements of fresh air demand under fast spatial-temporal variation of traffic conditions and operating environment. In this study, the operational data collected from an extra-long highway tunnel were deeply analyzed using big data technology. By combining traffic flow and environmental monitoring data, a data-driven perception model based on the Random Forests was structured. The prediction results show that the proposed model provides better performances as compared to contrast models, indicating it had better ability to adapt to the dynamic changes of in-tunnel operational status while realizing accurate prediction. The designed intelligent control strategies of ventilation facilities and traffic operation applying for different operational status would provide a theoretical basis and data support for lifting the level of intelligent control as well as promoting energy saving and consumption reducing in extra-long highway tunnels.