<p style='text-indent:20px;'>The evaluation of asphalt pavement structures has been a critical challenge in the field due to the practical limitations in methodology. In this paper, we propose a data-driven framework to evaluate structural performance of nineteen widely used asphalt structures in the Research Institute of Highway Ministry of Transport track (RIOHTrack). Specifically, we utilize the unsupervised machine learning method to delineate the similar and disparate performance among tested structures based on four years of falling weight deflectometer (FWD) experiments. Next, the structural performance is investigated on the temporal scale and the dynamic performance variations are captured over the course of the testing. Finally, experimental results are discussed and we provide essential evidence to aid future asphalt pavement design and construction.</p>