ABSTRACT

This research aims to evaluate and estimate the interface shear strength (ISS) in asphalt pavement using machine learning (ML) techniques. The study proposed the classification and regression trees (CART) model based on measured data collected from laboratory experiments. Three experimental parameters of testing temperature, normal load, and tack coat rate were considered as variables. The results showed that the developed CART model could explain more than 98% of the experimental data in a relatively short period. The testing temperature was found to have the most significant influence on the ISS, followed by normal load and tack coat dosage. Additionally, a parametric analysis of the interaction effects of input parameters on the ISS revealed that the higher test temperatures and lower normal loads reduced the ISS, while a high tack coat rate and low normal load corresponded to a lower ISS of the asphalt pavement.