Evaluating the Performance of Aluminum Oxide Nanoparticle-Modified Asphalt Binder and Modelling the Viscoelastic Properties by Using Artificial Neural Networks and Support Vector Machines

The effect of aluminum oxide nanoparticles (Al2O3) on the 60/70 penetration asphalt cement (AC) was investigated in terms physical and rheological characteristics by using Superpave testing procedures. Al2O3 at 3, 5, 7% concentrations were blended with grade AC. Conventional procedures adopted regarding characteristics, while dynamic shear rheometer (DSR) conducted to evaluate high low temperature failure parameters. In addition, heuristic modelling techniques, artificial neural networks (ANN), support vector machines (SVM) employed predict performance AC mechanical conditions. frequency sweep test multiple stress creep recovery (MSCR) results revealed that optimum composition 5% concentration considering since further addition resulted degradation enhanced properties due agglomeration blend. On contrary, demonstrated lowest viscoelastic behavior intermediate temperatures. higher complex modulus ( G ∗ ) lower phase angle id="M2"> δ parameters indicated increase stiffness modification process cost losing elastic against fatigue cracking. Moreover, based statistical indicator, coefficient determination (R2), it observed ANN models for predicting id="M3"> id="M4"> achieved a prediction accuracy 0.989 0.911 SVM able achieve 0.984 0.929, respectively, training datasets. other hand, noted outperformed smaller gap between obtained from difference datasets id="M5"> id="M6"> 3.2% 6.8% models, differences 11.6% 9.5%, indicating more prone overfitting phenomenon.