Early-Warning Model for Ice Formation on Environmentally Friendly Antifreeze Asphalt Pavement Using Support Vector Machines

Abstract

Adding antifreeze filler to the asphalt mixture can make the road surface have a certain automatic snow-melting function, but the snow-melting effect of antifreeze asphalt pavement (AFAP) is greatly affected by the environment. To ensure that road management can take other snow-melting measures in a timely manner when the road surface cannot melt snow, an early-warning model of AFAP was studied. To conduct a comparative analysis the freezing time under different environmental factors, ordinary asphalt pavement (OAP) and AFAP were used as subjects. The test results show that the temperature, humidity, and precipitation significantly impact the freezing time of pavement. A feature matrix of environmental factors was established by preprocessing the experimental data. After initially building the model and parameter tuning, an ice warning model for AFAP under multifactor conditions using support vector machines (SVMs) was constructed. Additionally, the model was used to predict the freezing time under different environmental factors. The prediction model based on the SVM could accurately predict the freezing time and has a high generalization ability. The research results provide a solution to early icing warnings for AFAP under specified scenarios.