A Novel Neural Computing Model Applied to Estimate the Dynamic Modulus (DM) of Asphalt Mixtures by the Improved Beetle Antennae Search

To accurately estimate the dynamic properties of asphalt mixtures to be used in Mechanistic-Empirical Pavement Design Guide (MEPDG), a novel neural computing model using improved beetle antennae search was developed. Asphalt were designed conventionally by eight types aggregate gradations and two binders. The modulus (DM) tests conducted under 3 temperatures loading frequencies construct 144 datasets for machine learning process. A network developed an (BAS) algorithm adjust hyperparameters more efficiently. predictive results proposed determined R RMSE importance score input parameters assessed as well. prediction performance showed that BAS can effectively calculation model, built mixture DM has higher reliability effectiveness than random hyperparameter selection. evaluate predict MEPDG. shear binder is most important parameter affects because its high correlation with adhesive effect composition. phase angle highest influence on remaining variables. these influences provide reference future design mixtures.