|
Title |
Development of Road Surface Condition Prediction Technology Using RWIS and Experimental Data
|
|
Authors |
노승지(Seung-Ji Noh) ; 이종한(Jong-Han Lee) |
|
DOI |
https://doi.org/10.11112/jksmi.2026.30.1.28 |
|
Keywords |
도로 노면 상태; 마찰계수; 랜덤 포레스트; 장단기 메모리; 예측 모델 Road surface condition; Friction coefficient; Random forest; Long short-term memory; Prediction model |
|
Abstract |
This study proposes a framework for predicting road surface friction using data collected from Road Weather Information System(RWIS).
The framework consists of two main stages: experimental verification of RWIS data reliability and development of a prediction model based on the verified data. Experimental comparisons verified that the friction coefficients measured by RWIS reliably reflect actual changes in road surface conditions, confirming their suitability as reliable indicators. Using the validated dataset, variable importance was evaluated through a Random Forest(RF), and the key variables identified were used to train a Long Short-Term Memory(LSTM) network that captures the temporal characteristics of road friction.
By integrating experimental validation with time-series prediction modeling, the proposed framework ensures both reliability and applicability in road surface condition prediction. The outcomes of this study provide a foundation for developing early warning systems for road icing and black ice, contributing to proactive winter road management and improved traffic safety.
|