Abstract
To explore the relationship between the asphalt properties and asphalt mixture properties, the data mining is conducted. The asphalt grade, optimum asphalt content, G∗, phase angle and G-R are selected as the asphalt indexes. The dynamic modulus, dynamic stability, crack progression rate and critical fracture energy are selected as the asphalt mixture performance indexes. Through the Pearson Correlation Coefficient analysis, all the asphalt indexes have the correlation with dynamic modulus; The asphalt grade, optimum ashphalt content and phase angle have the correlation with dynamic stability; Only optimum asphalt content has the correlation with the crack progression rate; The asphalt grade, phase angle and G-R have the correlation with the critical fracture energy. To establish more accurate models, the stepwise regression analysis is performed to screen the asphalt indexes. Through the analysis, the asphalt grade and optimum asphalt content are selected to predict the dynamic modulus; The optimum asphalt content and phase angle are selected to predict the crack progression rate and dynamic stability. The prediction models are established. However, the prediction model of critical fracture energy is imprecise and needs more research. Through the models, the optimum asphalt content range of high modulus asphalt concrete is recommended and the requirement of phase angle is determined.