講演情報

[AI-01]Machine Learning Based Forecasting of Winter Road Surface Grip

*Hao Wang1, Tianjie ZHANG1 (1. Rutgers University (USA))

キーワード:

Surface friction、Winter roadway maintenance、Artificial Intelligence、RWIS

Road surface friction is critical for traffic mobility and safety in winter weather. With road weather information system (RWIS) and artificial intelligence (AI), it is promising to develop data-driven models for effective decision-making in winter roadway maintenance winter maintenance. This study presents two case studies of using long-short term memory (LSTM) model with RWIS data collected during winter seasons to predict surface friction and determine salt application. The first case is to predict pavement surface grip values using time-series data collected from RWIS stations including climate variables and surface conditions. A balanced algorithm is introduced, effectively reducing data skewness and improving the model performance on low grip values. Analysis results demonstrate that the proposed RWIS-LSTM outperforms traditional machine learning models and other recurrent architectures. The second case is to predict road surface friction for salt applications at each snow event. It considers sequential effects of surface temperature, atmospheric condition, and salt application on the time-dependent evolution of road surface grip. The interaction effects of salt application and climate conditions on road surface friction with time were analyzed. Overall, this study demonstrates the potential of integrating AI and RWIS data to enhance winter roadway maintenance through improved prediction of road surface friction.