Title |
Development of Road Surface Condition Prediction Technology for Black-Ice Based on RF-LSTM Model
|
Authors |
노승지(Seung-Ji Noh) ; 안효준(Hyo-Joon An) ; 이종한(Jong-Han Lee) |
DOI |
https://doi.org/10.11112/jksmi.2025.29.3.20 |
Keywords |
블랙아이스 예측; 랜덤 포레스트; 장단기 메모리; 도로기상관측장비 Black-ice prediction; Random forest; Long Short-Term Memory; Road Weather Information System(RWIS) |
Abstract |
Black-ice is a major winter road hazard, forming a thin and nearly invisible layer of ice on the road surface that is difficult for drivers to detect. The risk of black-ice increases during conditions such as sudden temperature drops or cooling after snowfall, making early prediction crucial for road safety. However, existing prediction methods often fail to fully capture rapid weather changes and road surface condition, highlighting the need for more advanced predictive models. This study proposes a black-ice risk prediction model that combines a Random Forest(RF) algorithm with a Long Short-Term Memory(LSTM) network. The RF model is employed to identify key variables that are highly correlated with black-ice formation. These selected variables are used as inputs to the LSTM model, which captures temporal patterns relevant to freezing events. Evaluation results confirm that the proposed model effectively predicts black-ice risk. The developed model has potential for integration into real-time road safety management systems. It is expected to contribute to the prevention of traffic accidents caused by black-ice.
|